Some pages of this thesis may have been removed for copyright restrictions.
If you have discovered material in AURA which is unlawful e.g. breaches copyright, (either yours or that of a third party) or any other law, including but not limited to those relating to patent, trademark, confidentiality, data protection, obscenity, defamation, libel, then please read our Takedown Policy and contact the service immediately
RETINAL VESSEL ANALYSIS: FLICKER REPRODUCIBILITY, METHODOLOGICAL STANDARDISATIONS AND PRACTICAL LIMITATIONS
ANGELOS KALITZEOS Doctor of Philosophy
ASTON UNIVERSITY September
©Angelos Kalitzeos, Angelos Kalitzeos asserts his moral right to be identified as the author of this thesis
is copy of the thesis has been supplied on condition that anyone who consults it is understood to recognise that its copyright rests with its author and that no quotation from the thesis and no information derived from it may be published without proper acknowledgement.
Aston University
Retinal Vessel Analysis: Flicker Reproducibility, Methodological Standardisations and Practical Limitations Angelos Kalitzeos Doctor of Philosophy
esis Summary:
e Retinal Vessel Analyser (RVA) is a commercially available ophthalmoscopic instrument capable of acquiring vessel diameter fluctuations in real time and in high temporal resolution. Visual stimulation by means of flickering light is a unique exploration tool of neurovascular coupling in the human retina. Vessel reactivity as mediated by local vascular endothelial vasodilators and vasoconstrictors can be assessed non-invasively, in vivo. In brief, the work in this thesis • deals with interobserver and intraobserver reproducibility of the flicker responses in healthy volunteers • explains the superiority of individually analysed reactivity parameters over vendor- generated output • links in static retinal measures with dynamic ones • highlights practical limitations in the use of the RVA that may undermine its clinical usefulness • provides recommendations for standardising measurements in terms of vessel location and vessel segment length and • presents three case reports of essential hypertensives in a -year follow-up. Strict standardisation of measurement procedures is a necessity when utilising the RVA system. Agreement between research groups on implemented protocols needs to be met, before it could be considered a clinically useful tool in detecting or predicting microvascular dysfunction.
Keywords: Cardiovascular Disease, Autoregulation, Blood Pressure, Endothelial Function
Contents
List of Figures 7
List of Tables 9
1 Overview of principal literature 11 . e Heart ...... .. Cardiac Anatomy and Physiology ...... .. Conduction System ...... . Cardiovascular System ...... .. Blood Flow ...... .. Arterial Blood Pressure ...... .. Regulation of Blood Pressure ...... ... Mean Arterial Blood Pressure and Pulse Pressure ...... . Diagnostic Tests and Procedures ...... .. Ambulatory BP Monitoring ...... ... Ambulatory Arterial Stiffness Index ...... .. Electrocardiography ...... ... Ambulatory ECG Monitoring ...... ... Heart Rate Variability ...... .. CardioTens -hour ECG and BP monitoring ...... . Hypertension ...... .. Definition and Classification of Hypertension ...... .. Management of Hypertension ...... . Peripheral Circulation ...... . e Eye ...... .. Retinal Vasculature ...... .. Posterior Eye Haemodynamics ...... . alitative Retinal Analysis ...... .. e Keith-Wagener-Barker Classification ......
Contents
. antitative Retinal Analysis ...... .. Assessing Retinal Structure ...... ... Arterio-Venous Ratio ...... ... Vessel Tortuosity Index ...... ... Bifurcation Angles and Junction Exponents ...... ... Length-to-Diameter Ratio ...... .. Assessing Retinal Vessel Dynamics ...... ... Retinal Vessel Analyser ...... ... Flicker Provocation ...... ... Dynamic Response Analysis ...... ... Other External Provocations ...... .. Retinal Oximetry ...... .. Visual Field Testing ...... . Nailfold Capillaroscopy ...... .. Measurement Principle ...... .. CapiScope Capillaroscopy System ......
2 Retinal Vessel Analyser: Reproducibility 43 . Background ...... .. Motivation and Research Rationale ...... .. Aims ...... . Subjects and Methods ...... .. Interobserver Reproducibility ...... .. Intraobserver Reproducibility ...... .. Standardisations Applied ...... .. Inclusion and Exclusion Criteria ...... .. Study Protocol ...... ... Intraocular Pressure Measurement ...... ... Blood Pressure and Pulse Assessment ...... ... Retinal Vessel Functional Assessment ...... ... Retinal Vessel Structural Assessment ...... .. Outcome Measures ...... .. Dynamic Parameters ...... .. Static Parameters ...... .. Statistics and Data Analysis ......
Contents
. Results ...... .. Interobserver Reproducibility ...... ... Subjects ...... ... Retinal Vessels’ Absolute Diameters ...... ... Inbuilt Dynamic Flicker Response Analysis ...... ... Independent Dynamic Flicker Response Analysis ...... ... Comparison Between Inbuilt and Independent Analysis ... ... Averaged Flicker Responses ...... .. Intraobserver Reproducibility ...... ... Subjects ...... ... Retinal Absolute Diameters ...... ... Inbuilt Dynamic Flicker Response Analysis ...... ... Independent Dynamic Flicker Response Analysis ...... ... Comparison Between Inbuilt and Independent Analysis ... ... Averaged Flicker Responses ...... ... Static Retinal Vessel Parameters ...... ... Multiple Regression Analysis ...... . Discussion ...... .. Interobserver Reproducibility Study ...... .. Intraobserver Reproducibility Study ...... . Conclusions ......
3 Location and Length Influence on Vessel Reactivity 73 . Background ...... .. Motivation and Research Rationale ...... . Subjects and Methods ...... .. Data Collection ...... .. Data Processing ...... ... Analysis per Location ...... ... Analysis per Segment’s Length ...... .. Data and Statistical Analysis ...... . Results ...... .. Baseline Characteristics ...... .. Comparison Across Vessel Segments ...... .. Flicker Responses Variability as a Function of Location ...... .. Comparison Between Segment Lengths ......
Contents
. Discussion ...... .. Measurement Location Considerations ...... .. Measurement Length Considerations ...... . Conclusions ......
4 Essential Hypertension: Case Reports 92 . Introduction and Motivation ...... . Ethical Approval ...... . Methods and Subjects ...... .. Day - Ambulatory BP and ECG Monitoring ...... ... Outcome Measures ...... .. Day - Eye examinations ...... ... Intraocular Pressure Measurement ...... ... Retinal Functional Analysis ...... ... Outcome Measures ...... .. Data Analysis ...... . Results ...... .. Sample ...... .. hr BP and ECG Monitoring ...... .. Retinal Functional Assessment ...... . Discussion ......
Bibliography 99
Acronyms 114
Appendix 116
List of Figures
. Primary physiologic factors affecting certain physical factors...... . Graphical representation of the blood pressure wave...... . Two complete, normal ECG cycles...... . Modified three-electrode bipolar lead system...... . e time series of R-R intervals, i.e. tachogram...... . Power Spectral Density of the tachogram in Figure ....... . Set-up window of the CardioVisions soware...... . Normal fundus image of a le eye, ONH centered, °...... . Normal fundus image of a right eye, macula centered, °...... . Concentric AVR measurement rings as defined by the ARIC study...... . Schematic representation of the standard flicker measurement protocol. .... . A typical, normal arterial and venular flicker response......
. Typical measurement location...... . Measurement of the Arterio-Venous Ratio...... . Illustration of bifurcation angles measurement using ImageJ...... . Illustration of tortuosity measurement using ImageJ...... . Arteriolar diameter fluctuation and responses across all three flickers...... . Arteriolar reaction and constriction times across all three flickers...... . Venular diameter fluctuation and responses across all three flickers...... . Venular reaction times across all three flickers...... . Arteriolar diameter fluctuation and responses across all three flickers...... . Arteriolar reaction and constriction times across all three flickers...... . Venular diameter fluctuation and responses across all three flickers...... . Venular reaction times across all three flickers......
. RVA’s measuring window...... . Illustration of measured locations...... . Maximum Dilation Response and MABP Correlation......
List of Figures
. Comparison between segment length: Arteries...... . Comparison between segment length: Veins......
. years follow-up: Subject JH...... . years follow-up: Subject JW...... . years follow-up: Subject TR......
List of Tables
. Statistical measures of HRV used in time domain analysis...... . Geometrical measures of HRV used in time domain analysis...... . Classification of hypertension based on blood pressure levels...... . Keith-Wagener-Barker hypertension classification system...... . Vendor-supplied soware vessel response classification scheme......
. Sample demographics and baseline characteristics...... . Comparison of absolute arteriolar and venular diameters...... . Inbuilt Dynamic Flicker Response Analysis...... . Independent Dynamic Flicker Response Analysis...... . Reaction and Constriction Time Response Analysis...... . Inbuilt and Independent Analysis Comparison...... . Independent Dynamic Flicker Response Analysis...... . Reaction and Constriction Time Response Analysis...... . Sample baseline characteristics...... . Comparison of absolute arteriolar and venular diameters...... . Inbuilt Dynamic Flicker Response Analysis...... . Independent Dynamic Flicker Response Analysis...... . Reaction and Constriction Time Response Analysis...... . Inbuilt and Independent Flicker Analysis Comparison...... . Independent Dynamic Flicker Response Analysis...... . Reaction and Constriction Time Response Analysis...... . Static Retinal Vessel Parameters......
. Vessel Segments Absolute Diameters and Baseline Characteristics...... . Extended Vessel Segments Analysis: Flicker Responses...... . Extended Vessel Segments Analysis: Reaction Times...... . Extended Vessel Segments Analysis: Averaged Flicker Responses...... . Extended Vessel Segments Analysis: Averaged Reaction Times......
List of Tables
. Parameters’ Variability as a Function of Location...... . Comparison of Long and Short Segments...... . Diameter Deviation Between Long and Short Segments......
. Blood Pressure and Heart Rate Variability Parameters...... . RVA parameters across a five-year follow up......
Chapter
Overview of principal literature
. e Heart
.. Cardiac Anatomy and Physiology
e circulatory system of an adult individual contains a volume of approximately litres of blood (Rogers, ). e body organ responsible for its circulation to every tissue is the heart. Anatomically, the heart consists of four muscular chambers; the right and le atria (superiorly) and the right and le ventricles (inferiorly). ese are pairwise interconnected via two valves; the tricuspid and the mitral valve respectively. Exteriorly to the heart, these chambers are connected to the largest blood vessels of the human body; the vena cavæ (the superior and the inferior) and the aorta. Oxygen-deprived blood is fed to the heart via the vena cavæ, entering into the right atrium. Blood flow continues through the tricuspid valve into the right ventricle, while the valve opens and the atrium contracts (atrial systole). In turn, as soon as the right ventricle fills with blood, it contracts as well (ventricular systole). At the same time another valve, the pulmonary valve, opens up to let blood through to the pulmonary arteries to be pumped into the lungs. In the lungs, carbon dioxide is released and oxygen is absorbed. Now, oxygen-rich blood returns to the heart via the pulmonary veins and correspondingly passes through the le atrium, the mitral valve and into the le ventricle. is occurs at the same time as a new contraction is taking place in the heart’s right atrium-ventricle. e final valve that opens simultaneously with le ventricular contraction is the aortic valve. Here, blood is pumped into the aorta to be distributed further out towards the body’s organs and tissues.
.. Conduction System
e aforementioned muscle contractions are driven by a series of electrical impulses (action potentials) generated by a group of specialised cells in the right atrium, the cardiomyocytes. e tissue that comprises these cells is called the Sinoatrial (SA) node, also known as the
Overview of principal literature heart’s “natural pacemaker”. e conduction pathway of the heart is made up of the following specialised heart tissues: the Atrioventricular (AV) node, the bundle of His, the right and le bundle branches and the Purkinje fibres, successively. Finally, the actual contraction function is performed by the contractile cells of the heart. ese provide the necessary kinetic and potential energy for blood to propagate through the circulatory system.
Cardiac function is altered by neural activation. e heart is innervated by both the Sympathetic Nervous System (SNS) (adrenergic) and the Parasympathetic Nervous System (PNS) (cholinergic). e two systems work in tandem; the former is known to be the accelerator of the heart, whereas the laer serves as the heart’s decelerator. Epinephrine and norepinephrine are the two main chemical mediators that increase Heart Rate (HR), AV conduction and contractility, via the SNS. e overall effect of sympathetic contribution is to increase Cardiac Output (CO), Systemic Vascular Resistance (SVR) and arterial Blood Pressure (BP). e time necessary for the SNS to actuate these effects is in the order of seconds (Clifford et al., ). is compensating mechanism is particularly important during exercise, emotional stress and haemorrhagic shock. On the other hand, parasympathetic innervation of the heart is controlled by the vagus nerve. Parasympathetic activity therefore is sometimes termed vagal activity. In contrast with the SNS, the vagus nerve acts quickly, carrying impulses that lower HR and decelerate or block AV conduction, via acetylcholine release, within a second.
. Cardiovascular System
.. Blood Flow
e flow of fluids with viscosity (η) through rigid, cylindric tubes of length (L) and radius (r) in hydraulic systems obey the Poiseuille-Hagen law (Pournaras et al., ) that expresses the relation between the fluid flow (Q) and the pressure difference (or else perfusion pressure) (ΔP):
ΔP ﻀ ΔPπr Q = = (.) ηL Rﻄ
e above expression assumes long, straight tubes, a Newtonian fluid and steady, laminar flow conditions. Despite the fact that these assumptions are not entirely true for the human vascular system, the flow, pressure and resistance relationships still remain applicable. Hence, in analogy to electrical circuits and Ohm’s law, the rate of blood flow (Q) is inversely proportional to vascular resistance (R). From Equation (.):
Overview of principal literature
ηL R ∝ (.) ﻀ r In health, blood viscosity does not fluctuate, assuming constant haematocrit and temperature, thus it can be considered constant (Klabunde, ). Similarly, vessel length is constant. erefore, the major determinant of resistance to blood flow through a blood vessel is its calibre .((Equation (.) ﻀ r/ﺽ diameter is directly proportional to radius), since it is proportional to) is means that a potential halving of blood vessel radius increases resistance -fold and vice versa. It is evident that subtle vessel calibre fluctuations can yield marked changes in vascular resistance and consequently in blood flow.
μm), arterioles are termed resistance vessels andﺼﺼﺾ Due to their small diameter (i.e. less than may regulate local blood flow in their surrounding tissues (Klabunde, ). e extrinsic regulation is mediated via a twofold pathway; the Autonomic Nervous System (ANS) and the endocrine system. At rest, arterioles receive a baseline level of autonomic stimulation which makes them slightly constricted, known as the vascular tone. Vasodilation is reached by decrease of sympathetic stimulation below baseline levels and vasoconstriction by increase above the baseline. Other mechanisms controlling vasomotor response, intrinsic to the vessels this time, take the form of metabolic and myogenic control; the former is incurred by metabolite accumulation according to the rate of metabolic activity and the laer by smooth muscle relaxation or contraction. Metabolic regulation is mainly mediated by vascular endothelial cells and local neural tissue surrounding the vessels, which release vasoactive molecules. e most potent vasodilator is known to be Nitric Oxide (NO) and conversely endothelin- is the most potent vasoconstrictor (Haefliger et al., ). e myogenic responses for blood flow vascular autoregulation are mediated by pericytes and smooth muscle cells. Myogenic vascular tone is a function of artery wall stretch and depends on the presence of calcium in the extracellular space (Bevan et al., ).
.. Arterial Blood Pressure
e arterial vascular network is considered to be physically determined by its elastic char- acteristics (compliance) and the blood volume circulating in it. e arterial volume in turn depends on the inflow rate from the heart into the arteries (CO) and the outflow rate from the arteries through the resistance vessels (arterioles) into tissues and body organs (Figure .). In the resting state, these peripheral organs are supplied with blood according to their metabolic needs. From the law of conservation of mass, given that the vascular system is a closed-loop circuit, if the heart pumps blood at a higher rate than it is fed through veins, then the arterial walls need to expand, giving a pressure rise and vice versa.
Overview of principal literature
Physiologic factors
Cardiac output (Heart rate Physical factors × Stroke volume) Arterial blood volume Arterial blood pressure Peripheral Arterial compliance resistance
Figure .: Primary physiologic factors affecting certain physical factors which, in turn, determine arterial blood pressure. Adapted from Berne and Levy () (p. )
Arterial BP is the force of blood against artery walls due to the pumping of the heart. BP is defined by the two extremes of systemic BP, namely systolic and diastolic pressure. Systolic Blood Pressure (SBP) is the highest measured value obtained during ventricular contraction (during a heartbeat), whereas Diastolic Blood Pressure (DBP) is the lowest measured value obtained during ventricular relaxation (between heartbeats). BP is measured in millimeters of mercury (mmHg). Automatic, digital BP devices using the oscillometric measurement method are clinically used. Measurement is performed by occluding the artery of an extremity (arm, wrist, finger, or leg) with an inflatable cuff (Beevers et al., ).
.. Regulation of Blood Pressure
In physiologic conditions, whenever the sympathetic system is activated, the body down- regulates parasympathetic activity and vice versa. ese two branches of the ANS are rarely completely activated or deactivated; instead the body adjusts their levels of activation appropriately to its needs. On that basis, the ANS makes HR adjustments via sensors located throughout the body. ese sensors include the baroreceptors which exist in all mammalian arteries and sense the arterial pressure. ey are a type of mechanoreceptor and respond to arterial wall stretching, in a negative feedback loop fashion. If arterial pressure (mean, pulse or both, see Section ...) rises abruptly, then vessel walls passively dilate in order to accommodate this pressure rise. Consequently, the baroreceptors get activated and fire action potentials to a degree proportional to the change in pressure. e baroreceptor firing has an inhibitory effect on sympathetic outflow and a boosting effect on vagal outflow, hence driving BP down. Also, these autonomic changes cause vasoconstriction (increased SVR) and decreased CO. e reduction in CO results from both a decreased HR and a reduced force of AV contraction, dropping arterial BP. e reverse action is taking place in case of a sudden pressure drop, sustaining arterial BP at normal levels at all times. is is known as the baroreceptor
Overview of principal literature reflex or simply baroreflex, one of the body’s mechanisms capable of maintaining homoeostasis. Baroreflex sensitivity is now a prognostic factor in cardiology; it is significantly altered during certain disease states (La Rovere et al., ).
... Mean Arterial Blood Pressure and Pulse Pressure
When BP is clinically measured, the systolic and diastolic values are recorded. But two additional ways of characterizing BP are important to consider. Physiologically, the pressure that is primarily regulated and considered a beer indicator of perfusion (than SBP) to vital organs, is the Mean Arterial Blood Pressure (MABP). MABP, measured in mmHg, is the pressure in the arteries averaged over a single cardiac cycle duration and can be approximated from the following empirical formula:
ﺽ ﺾ MABP ≃ DBP + SBP (.) ﺿ ﺿ
Since the heart spends more time in the relaxing state (diastole) than in the contracting state (systole), DBP has a greater effect on MABP. For example, if systolic pressure is mmHg and diastolic pressure is mmHg, then MABP is approximately mmHg using the above calculation. A MABP of at least mmHg is necessary to perfuse the coronary arteries, brain, and kidneys. Factors that determine MABP are CO and SVR, according to the following relationship (based on Equation (.)) :
MABP ≃ CO ⋅ SVR (.)
As the le ventricle ejects blood into the aorta, the aortic pressure increases. e maximal change in aortic pressure during systole represents the aortic pulse pressure. Pulse Pressure (PP) is defined as the difference between arterial systolic and diastolic pressures. us, Equation (.) can be rewrien as:
ﺽ MABP ≃ DBP + PP (.) ﺿ
Using the same numeric example as previously (SBP/DBP equal to / mmHg) then PP equals to mmHg. e rise in aortic pressure from its diastolic to systolic value is determined by the compliance of the aorta as well as the ventricular Stroke Volume (SV). e stroke volume is the amount of blood injected into arteries by each heart beat:
Overview of principal literature
SBP 120 dicrotic notch
pulse pressure MABP Pressure (mmHg)
80 DBP
0 0 0.8 Time (sec)
Figure .: Graphical representation of the BP wave. SBP - DBP = PP. e dicrotic notch is caused by the aortic valve closure. MABP is defined according to Equation (.). Duration of a cardiac cycle is approximately msec.
CO SV = (.) HR e greater the stroke volume the heart pumps out, the greater the change in aortic pressure. Compliance is simply a measure of the capacity of the arterial system to accommodate further increases in volume (Δvolume / Δpressure). At given stroke volumes, the pressure increase is determined by the vessel compliance. Flexible arteries that expand easily have high compliance, contrary to stiff arteries. e aorta’s walls, being the most compliant vessel walls throughout the arterial system, expand to accommodate the increase in blood volume with ventricular ejection. e more compliant a vessel, the smaller the pressure change during cardiac cycles (i.e. smaller PP). Aortic compliance decreases with age or disease (e.g. arteriosclerosis) due to structural changes, thereby producing age-dependent increases in PP. In general, both an abnormally high PP and MABP are risk factors for cardiovascular disease (Dart and Kingwell, ).
. Diagnostic Tests and Procedures
.. Ambulatory BP Monitoring
Since BP is depending on factors such as age, gender, time of day (Millar-Craig et al., ), diet, cardiac cycle (Knudtson et al., ; Chen et al., ), stress and physical activity (Anuradha et al., ) it is evident that isolated, clinic BP measurements have inherent limitations and may not be representative of the true BP in many patients. e white-coat effect, referring to a BP increase occurring at the time of a clinic visit and aenuating soon thereaer, may be an additional obstacle to true BP assessment (Pickering et al., ). An alternative
Overview of principal literature method of measuring BP in clinical practice is self-monitoring or home monitoring. Multiple measurements at home enable a beer estimate of the average or true pressure, it is cost- effective and usually eliminates the white-coat effect (Verberk et al., ). Nevertheless, it is prone to measurement bias due to unreliable procedures when obtaining BP.
Portable BP monitors that can be fied for durations of to hours and can obtain and record regular BP readings have become widely accepted as a clinically useful tool for diagnosing and managing Hypertension (HT)(Waeber et al., ). Rather than measuring BP when patients are remaining under artificial clinic conditions, BP behaviour is recorded during their usual daily activities. Ambulatory Blood Pressure Monitoring (ABPM) offers a wealth of BP related information that no other method can provide. Mapping of diurnal variation of BP, calculating arterial stiffness indices and pinpointing transient BP events would not be achieved without the advent of ABPM. Measurement frequency during a typical -h period is generally not recommended to be greater than every minutes (which could interfere with routine activities), neither less frequent than every minutes (which could give an inadequate amount of measurements). Pieces of evidence from cardiovascular event-based, longitudinal studies have been made available, that ABPM improves cardiovascular risk stratification over and beyond traditional risk factors, including conventional clinic BP measurement (Verdecchia, ).
... Ambulatory Arterial Stiffness Index
Another surrogate marker of arterial stiffness derived from ABPM that may predict cardiovas- cular mortality is the Ambulatory Arterial Stiffness Index (AASI)(Dolan et al., ). AASI is derived graphically by ploing DBP against SBP readings from unedited -hour recordings and obtaining the slope of the regression line. AASI is then calculated as one minus the regression slope. It is a novel measure which has been shown to be an integrated measure, which is characteristic for an individual and reflects the combined effects of le ventricular ejection, active and passive components of arterial stiffness, and the reflection of the arterial pulse wave (Li et al., ). Prior to the introduction of AASI, researchers were using PP as a measure of arterial stiffness. However, PP only reflects a static difference between systolic and diastolic pressure and does not exploit the dynamic relation between diastolic and systolic blood pressure throughout the whole day as AASI does.
.. Electrocardiography
e electrical activity of the heart can be monitored in a non-invasive manner by means of an electrocardiograph. Changes occurring in the membrane potential of cardiac muscle cells during consecutive cardiac cycles are added up to produce an Electrocardiogram (ECG). ECG
Overview of principal literature recordings provide a plethora of information about cardiac structure and function over time. It is widely used in clinical practice to diagnose heart disorders including cardiac arrhythmias. A number of electrodes are aached to the skin at pre-specified positions sensing the electrical currents which propagate from the heart’s surrounding tissue to the skin surface. ese weak, at first, electrical currents are transmied to an ECG device, amplified and transformed into ECG traces that represent the heart’s systole-diastole cycle. A schematic representation of an ECG tracing is shown in Figure ..
Cardiac cells at rest are considered polarised. When a stimulus occurs (SA node firing), ions cross the cell membrane and cause an action potential resulting in atrial contraction. is is called depolarisation and is represented by the P wave. e electrical impulse arrives at the AV node where it is delayed by milliseconds. Albeit minute, this delay is critical: it allows the atria to fully contract and eject blood into the ventricle. At the same time, it keeps the ventricle from contracting too quickly, allowing adequate time to complete its filling phase. On the ECG, this brief period of no electrical activity is depicted by a straight (isoelectric) line between the P wave and the beginning of the QRS complex. e impulse is then propagated down to the ventricle through the bundle of His, right and le bundle branches and Purkinje fibres yielding ventricular contraction (QRS complex). Finally, the ventricle returns to its relaxed state (T wave); this is called repolarisation. HR can be determined from the R-R interval, which is the time between consecutive QRS complexes. Here, the term HR refers to the rate of ventricular contractions. In some abnormal conditions atrial and ventricular rates differ, so it is important to distinguish between the two. Atrial HR is determined by measuring the P-P intervals.
Obtaining an ECG waveform is possible by means of one or more leads simultaneously. A lead provides a view of the heart’s electrical activity between a positive and a negative pole. Intuitively, these are called bipolar. Connecting these two poles by an imaginary line defines the lead’s axis, which refers to the direction of the electrical current flowing through the heart. e ECG output consists of an upward deflection if electrical current is heading towards the (+) electrode and vice versa. All bipolar leads need a third electrode as well, known as the ground. is is placed on the sternum bone to prevent electrical interference from reducing the ECG signal’s quality. Electrode terminals are color-coded for easier identification and placement.
Since the heart is a three-dimensional organ and ECG electrodes may only be placed super- ficially on the skin, there are two different planes that electrical activity can be probed from. ese are the frontal and the horizontal planes. Conventionally, for the frontal plane there are six limp leads, bipolar (I, II, III) and unipolar (avR, avL, avF) and for the horizontal plane
Different .(ﻂto V ﺽthere are six chest (alternatively called precordial) leads, all unipolar (V leads provide different diagnostic information. Unipolar leads require only one electrode.
Overview of principal literature
RR interval R R
PR segment
ST P segment T P T
PR Q Q interval S QT S interval QRS interval
Figure .: Two complete, normal ECG cycles with waves, intervals and segments shown.
Combining more than one lead and omiing others is normal clinical practice on a per patient basis.
An ECG may be recorded for a short time period (for example, minutes), so the output can be manually processed from hard copy printouts of the ECG strip created. Alternatively, longer recordings ( or hour) demand digital storage on solid state disks and specialised soware for analysis and interpretation.
... Ambulatory ECG Monitoring
Norman J. Holter, an American biophysicist, developed the first clinical prototype of a portable ECG recorder in (Barold, ). Since then, the term Holter monitoring is a synonym of Ambulatory Electrocardiography (AECG). ere are two types of AECG monitors. Continuous ones that typically record for or hours and intermient ones that are used for long periods of time (weeks to months) to provide shorter, intermient recordings on demand (alternatively termed event recorders).
Ambulatory cardiac monitoring overcomes the limitations of one-off ECG recordings as it is designed to identify transient cardiac disturbances occurring throughout the patient’s daily routine, thus deemed free of bias of a controlled laboratory seing. It also makes possible the examination of cardiac autonomic function by measuring Heart Rate Variability (HRV) (see Section ...). According to the Seventh Report of the Joint National Commiee clinical situations in which AECG monitoring may be useful are white-coat hypertension, evaluation of nocturnal BP changes and hypotensive symptoms associated with antihypertensive med- ications or autonomic dysfunction. Modern AECG monitors have an autonomy of storage
Overview of principal literature capacity and running time of digitally recording more than . QRS complexes (for a h period) and at the same time are compact and lightweight to carry on. Modified three-electrode bipolar leads have been developed for the case of AECG. Some of these electrode placements are shown in Figure ..
All modified bipolar lead placements shown in Figure . have the plus electrode in position
(ﻁat is, the fih intercostal space at the le anterior axillary line. e Central Back (CB .ﻁV lead (ﻁlead has the negative electrode at the right scapula bone. e Central Manubrium (CM lead has the negative electrode ﻁhas the negative lead at the manubrium sterni. Lastly, the CC
Rﻁat the fih intercostal space at the right anterior axillary line (alternatively, this is the V position). Typically, combinations of two or three bipolar leads are used, which amounts to five or seven electrodes in total, respectively. Signals are recorded in separate channels per electrode pair.
e aforementioned modified ambulatory leads offer maximised P-wave height for the dia- gnosis of atrial arrhythmias and increased ECG sensitivity for the detection of anterior lead as being ﻁmyocardial ischaemia. Particularly, a study has exemplified the use of the CM the most useful one for ambulatory monitoring (yyumi et al., ).
... Heart Rate Variability
e balance between the two reciprocal activities of the ANS (SNS and PNS) is evidenced in the beat-to-beat changes of the cardiac cycle. HRV is concerned with the oscillation (i.e. variability) in the interval between consecutive heart beats, which may contain indicators of current disease, or signs about impending cardiovascular disease. It can be quantitatively and non-invasively evaluated either by time domain or by frequency domain methods (Malliani, ). Comparisons of the sympatho-vagal balance can be made between pathological and physiological conditions, different types of activity (rest, exercise) and to analyse circadian rhythms (day-night changes).
Ground electrode is not shown. LA .ﻁ﹔﹔ ,ﻁ﹞﹔ ,ﻁFigure .: Modified three-electrode bipolar lead system: ﹔ stands for Le Arm, RA stands for Right Arm.
Overview of principal literature
Figure .: e time series of R-R intervals obtained is represented graphically as a tachogram. e horizontal axis of the diagram indicates the time and the vertical axis shows the RR distance in msec (average of seconds). One data line shows the minimum and maximum RR intervals. See also Figure ..
In case of time domain HRV analysis, the so-called Normal-to-Normal (NN) intervals are determined. ese are all intervals between normal (sinus) beats. Assuming a -hour AECG recording, the variables that can be calculated to estimate an overall HRV are either of statistical nature (Table .) or of geometrical nature (Table .). e list of variables in the aforementioned tables is not exhaustive (Malik et al., ). SDNN estimates (and similarly other HRV measures) depend on the length of the recording period. As such, valid comparisons are only to be carried out between values derived from ECG recordings of comparable durations.
e series of NN intervals can also be ploed accordingly to derive useful clinical correlates. e geometrical time domain methods are derived from various approaches implemented to characterize the variability of these plot paerns. To perform such calculations, either the sample density distribution of NN intervals is constructed (assigning the number of equally long NN intervals to each value of their duration) or a (D or D) Lorenz/Poincaré plot of NN intervals (each NN interval is ploed against its next one) (Hnatkova et al., ). Such geometrical methods require the NN intervals sequence to be appropriately binned on their time scale to permit the construction of smoothed histograms. e reason that these bins are
Variable Units Description SDNN msec Standard Deviation of all NN intervals Standard Deviation of the Averages of NN intervals SDANN msec calculated over minutes ECG segments (Square) Root of the Mean of the Sum of the Squares of RMSSD msec Differences between successive NN intervals
Table .: Statistical measures of HRV used in time domain analysis.
Overview of principal literature selected to be of approximately milliseconds (see Table .) is simply because it corresponds to the typical sampling frequency of commercial AECG monitors ( Hz).
Taking a step further from time domain into frequency domain analysis, spectral information from an R-R tachogram can be decomposed and periodicities may be identified. Power spectra of R-R variability from AECG recorders have been shown to provide markers of sympathetic and vagal function. In humans, three main spectral components are distinguished in a Power Spectral Density (PSD) plot (Figure .). e Very Low Frequency (VLF) component (. - . Hz) that depends primarily on the presence of parasympathetic outflow (Taylor et al., ). e Low Frequency (LF) component (. - . Hz) which is believed to be due to baroreceptor mediated BP control and relates to both sympathetic and parasympathetic function and the High Frequency (HF) component (. - . Hz) which is correlated with respiratory driven vagal input to the SA node, reflecting parasympathetic nervous system activity (Kamath and Fallen, ). Measurement of VLF, LF and HF power components is made in absolute values of Alternatively, LF and HF specifically may also be reported in normalised units .(ﺾ power (msec (nu) which represent the relative value of each component in proportion to the total power minus the VLF component (Pagani et al., ). e LF/HF ratio is an index of sympatho-vagal balance (Lombardi et al., ). High values for the ratio suggest predominance of sympathetic nervous activity and vice versa.
In hour AECG recordings of normal subjects, despite the fact that LF and HF components account for approximately only % of the total spectral power (see Figure .), these two are the spectral components mostly referred to in literature. LF and HF can increase under different conditions. In healthy subjects, an increased LF is observed during ° head tilt, standing, mental stress and moderate exercise, whereas increased HF is induced by controlled respiration and cold stimulation of the face (Malik et al., ).
e role of the ANS in essential HT (see Section .) is an important area of investigation. A sig- nificant amount of studies have investigated the clinical value of HRV in various cardiovascular
Variable Units Description HRV Total number of all NN intervals divided by the maximum triangular n/a height of the histogram of all NN intervals measured on a index discrete scale of . msec bins (/ sec) Width of the base of the triangular interpolation (the minimum square difference is used to find such a triangle) TINN msec of the maximum height of the histogram of all NN intervals
Table .: Geometrical measures of HRV used in time domain analysis.
Overview of principal literature
Figure .: Power Spectral Density of the tachogram in Figure .. Horizontal axis represents frequency in Hz and the vertical axis shows the PSD. e input range, total power, LF and HF power components and LF/HF ratio are displayed, too. e grey vertical lines on the chart mark the VLF, LF and HF components (-. Hz, .-. Hz, .-. Hz, respectively). diseases, including HT. Among those studies, an increased LF component during night-time rest has been found in hypertensives compared to normotensives, accompanied by blunting of circadian paerns (Guzzei et al., ). Complementary evidence of reduced parasympathetic cardiac control was found a few years later between a group of hypertensives and both normal and borderline hypertensive groups (Langewitz et al., ).
.. CardioTens -hour ECG and BP monitoring
CardioTens (Meditech Ltd, Hungary) is a commercially available combined ambulatory BP and ECG monitor. e device can be used either as an independent ABPM device if only the BP cuff is aached to it, or as an AECG if only the ECG electrodes are aached or as a dual recorder incorporating both functions simultaneously. e incorporated ABPM device is validated by the British Hypertension Society and performs within the recommendations of the Association for the Advancement of Medical Instrumentation (Barna et al., ). It uses a proprietary Meditech algorithm for determining BP which is equivalent to that obtained by a trained observer using the cuff/stethoscope auscultation method Korotkoff phase V, within the limits prescribed by the American National Standard for Electronic or Automated Sphygmomanometers (White et al., ). Readings can be taken at frequent, pre-programmed time intervals.
e AECG part of the device can perform routine AECG registration via two independent of seconds duration (ﻁand CC ﻁchannels, producing an ECG strip with two leads (CM every minutes, continuously for up to hours. It features a sampling rate of Hz, adequate to precisely locate the peak of QRS complexes, and an analog-to-digital converter
Overview of principal literature
Figure .: Set-up window of the CardioVisions soware for confirming correct electrode placement and checking
(channel B (yellow, green ,ﻁ﹔﹔ ECG signal in real-time. Channel A (red, white) corresponds to lead .and N electrode is the ground ﻁ﹞﹔ corresponds to of bits. PSD processing in CardioVisions uses four-minute Hann-windowed samples. If the input range is longer than this four-minute unit, PSD is calculated by averaging adjacent, non-overlapping four-minute samples. Individual spectral components can be automatically determined, together with their center frequency and associated power, i.e. area.
. Hypertension
.. Definition and Classification of Hypertension
Diagnosis and treatment of hypertension is essentially based on the outcome of casual, indirect BP readings, although ambulatory or BP recordings at home might eventually become preferable to minimise the bias of white-coat HT. A systolic and/or diastolic BP measurement persistently higher than normal values - for a given age group - is defined as HT. According to the latest and most relevant (for the European population) guidelines jointly published (Mancia et al., ) by the European Society of Hypertension and the European Society of Cardiology, normotensives and hypertensives are distinguished as in Table ..
Category Systolic BP (mmHg) Diastolic BP (mmHg) Optimal < and < Normal - and/or - High normal (Pre-Hypertension) - and/or - Grade HT (mild) - and/or - Grade HT (moderate) - and/or - Grade HT (severe) ≥ and/or ≥
Table .: Classification of HT based on BP levels in adults (Mancia et al., ). In case of SBP and DBP falling into different categories, the higher value is considered for classification.
Overview of principal literature
HT diagnosed in the vast majority of the population is of an unknown cause (i.e. idiopathic). is condition is called primary or essential HT. In both developed and developing countries, essential hypertension affects –% of the adult population, and up to –% of those beyond their seventh decade of life (Staessen et al., ). HT in the rest of hypertensive patients, results secondarily from renal disease, endocrine disorders, or other identifiable causes and this type is called secondary hypertension.
Malignant Hypertension (MHT) is the most severe form of HT, and is defined clinically as the presence of severe hypertension in association with ocular disease (Shantsila et al., ).
.. Management of Hypertension
Guidelines have been published from the World Health Organisation and the International Society of Hypertension on management of HT (Whitworth, ). More recently, the British Hypertension Society working party, in light of the latest peer-reviewed publications, has also published relevant guidelines (Williams et al., ).
Non-pharmacological strategies can reduce BP (Williams et al., ). ese entail reduction of alcohol consumption, lowering of salt intake, adopting a diet rich in fresh fruits and vegetables and restriction of caloric intake. Less effective for BP reduction but helpful also towards reducing cardiovascular risk in general, are regular dynamic exercise and abstaining from smoking.
Antihypertensive drug treatment diminishes the complications of HT. e three broad classes of drugs used to treat primary HT are diuretics (to reduce blood volume), vasodilators (to decrease SVR), and cardioinhibitory drugs (to decrease CO). Irrespective of the mechanisms that may operate to initiate and sustain HT, its treatment is important because it increases the risk of further complications such as coronary artery disease, stroke and renal disease.
. Peripheral Circulation
Although the site of routine BP measurement focuses aention on the haemodynamics of large conduit arteries, namely the brachial artery from the upper arm, it is accepted that HT is a systemic condition involving the vascular tree as a whole. As such, various non-invasive clinical tools have been developed to assess microvascular function and structure. Described in the following sections are two such vascular beds that can be affected by generalised systemic disturbances or ,in fact, be the ones that undergo pathological changes prior to systemic disease: the retinal circulation and the nailfold microcirculation.
Overview of principal literature
. e Eye
e eye is an easily accessible, transparent “window” to peripheral microvasculature that can be examined non-invasively in vivo. Two vascular beds exist in the posterior eye; the choroidal (part of the uveal layer) and the retinal. On one hand, retinal blood vessels supply nutrients and oxygen to the neural retina, namely retinal ganglion cells and their axons, as well as to the anterior part of the Optic Nerve Head (ONH). On the other hand, the choroidal plexus, is the most perfused tissue of any other tissue in the body per unit weight (Nickla and Wallman, ). Hence, the maintenance of normal fundus vascular structure and function is of high importance. In the following sections, retinal circulation features are described along with current clinical and laboratory investigative techniques on retinal structure and function.
.. Retinal Vasculature
e main vessels of the eye comprise the Central Retinal Artery (CRA) and the Central Retinal Vein (CRV). ese enter and exit the globe respectively within the ONH, bifurcate at the optic disk into superior and inferior branches and then further divide into temporal and nasal branches (Figure .). Mapping the retina into four quadrants yields the respective superior temporal, superior nasal, inferior temporal and inferior nasal vessel branches. e temporal parts of these vessel branches form the superior arcade, composed of the superior temporal artery and vein and the inferior arcade, composed of the inferior temporal artery and vein. In terms of fundus photography the temporal half of the retina is the target that is documented, including the ONH and the macula (Figure .). Stokoe and Turner (Stokoe and Turner, ) reported that the temporal side of the fundus contained wider vessels and were more predictable in their branching than the nasal one, when they were trying to obtain comparable vessel pairs (arteries and veins) for their study.
Figure .: Normal fundus image of a le eye, Figure .: Normal fundus image of a right ONH centered, °. eye, macula centered, °.
Overview of principal literature
Whilst termed retinal arteries and veins, retinal vessels aer they bifurcate for the first time distal to the ONH are in fact arterioles and venules respectively, if accurate terminology is used. Namely, the CRA begins to change markedly in its structure aer passing through the lamina cribrosa of the sclera, losing a big part of its internal elastic lamina and its muscular coat (Scheie, ). Aer the first bifurcation distal to the ONH, the arterioles and venules have no elastic lamina and the muscle fibers lose continuity. en, arterioles gradually branch off to form smaller arteriole daughter vessels and terminal arterioles, which feed into the capillary bed as they extend towards the peripheral retina. Pre-capillary arterioles and post-capillary venules are linked through anastomotic capillaries.
.. Posterior Eye Haemodynamics
Blood circulation of the posterior eye and especially choroidal blood flow depend on perfusion pressure. Mean Ocular Perfusion Pressure (OPP) driving blood through the eye is the mean blood pressure in the ophthalmic artery entering the orbit minus the pressure in the veins returning to the heart. e venous pressure is approximately equal to Intraocular Pressure (IOP), while there is a pressure drop of a factor of / between the brachial artery and the ophthalmic artery (Pournaras et al., ), thus:
ﺾ meanOPP ≃ MABP − IOP (.) ﺿ
Retinal blood vessels are not innervated; rather their dilation and constriction depend on autoregulation (Dorion, ). e main regulators of retinal blood flow are the vascular endothelium cells, the neural and the glial cells. Experimentally, autoregulation of the retinal microcirculation is assessed by provocation methods, which are extensively described in Sections ... to ....
e choroid, in contrast, does not exhibit autoregulation of blood flow. is, however, does not mean that the choroid is a passive, non-reactive vascular region. An intensive autonomous innervation by the SNS permits central regulation of the choroidal blood flow. is is important, for example, for protecting the choroid from hyperperfusion in patients with increased BP.
. alitative Retinal Analysis
e quest for early detection of systemic vascular disease by means of ophthalmologic examination has surely been long-lasting and is still ongoing. One of the earliest reports “on ophthalmoscopic evidence of general arterial disease” dates back to (Gunn, ).
Overview of principal literature
Markus Gunn first precisely defined a number of signs of the retinal vessels that are typical of retinal arteriosclerosis and demonstrated their close relation to cerebral vascular disease. Specifically, these signs have been further investigated by Moore (Moore, ) and comprised of a) irregularity of the lumen of retinal arteries, b) arterial tortuosity, c) increased arteriolar light reflex, d) loss of arterial wall translucency, e) venular blood flow obstruction where they are crossed by arteries (this condition was later termed as “arteriovenous nipping”) and ) retinal oedema.
.. e Keith-Wagener-Barker Classification
By early ’s, investigators had identified two distinct types of essential HT; the benign and the malignant form of the disease. However, it was apparent that this grouping was rather crude and did not facilitate all cases. In , Keith et al. () aempted to relate retinal vascular changes to survival rates in the hypertensive population aiming towards increased accuracy of prognosis in the general population. To avoid descriptive terms that could cause confusion, they used numbered groups. e so-called Keith-Wagener-Barker classification system appears in Table ..
eir results were based on identifying a combination of structural changes: qualitative retinal observations and quantitative measurements of peripheral arterioles of the pectoralis major muscle. e hypertensive patients included in their study were followed up for a period of to years and were grouped on the basis of the ophthalmoscopic characteristics of each group (Table .). e resulting survival curves were distinct for each of the four groups having a gradually increasing steepness from benign to malignant hypertension. Despite the inherent limitations of their study, described not only by later publications (Chasis, ) but by the authors themselves too, the Keith-Wagener-Barker classification scheme has been widely adapted (Walsh, ) in prognosticating for survival.
Table .: Keith-Wagener-Barker hypertension classification system (Walsh, ).
Overview of principal literature
. antitative Retinal Analysis
.. Assessing Retinal Structure
Digital retinal images are taken by means of fundus cameras. ese images can be post- processed in order to yield a breadth of quantitative values towards the ultimate goal of vascular network characterisation and risk stratification. ese quantitative metrics are discussed below.
... Central Retinal Arteriolar, Venular Equivalent and Arterio-Venous Ratio
e arteriole and venule widths and their ratio have long been regarded as signs of hypertensive disease (Kagan et al., ). In , Parr and Spears (Parr and Spears, a,b) paved the way towards the long-sought transition from qualitative and subjective grading of retinal images to quantitative and objective measurements by summarising the calibre of all retinal arteries as the equivalent width of the CRA. at way, comparison of the arterial widths of different eyes was made possible, independently of the complexity and paern of branching. ey recruited normotensive young adults and measured the diameter of all arterioles (parent-daughter pairs) from the edge of the ONH outward to about °. e rationale behind this zone selection is that at that distance from the optic disk the retinal arteries and veins are rather arterioles and venules respectively (i.e. they have lost their internal elastic lamina and their muscle layer is not continuous (Scheie, )) and according to studies (Parr, ) it has been suggested that these vessels are more readily affected from pathological conditions, like HT. en, they investigated the relationship between individual trunk vessels and their corresponding branches and calculated a model that best fit their experimental data. Next, they confirmed their model with an independent group of subjects. is empirically derived formula calculated the width (in μm) of a parent artery from the widths of its two branches:
ﺾ ﺾ (.) (ﻂﻃ.ﺼﺽ − ﺾWﺽWﺾﺾ.ﺼ − ﺾWﺽﺼ.ﺽ + ﺽWﻃﻄ.ﺼ)W︀︑︓︄︑︘ = ｮ the wider ﺾthe narrower and W ﺽwhere W︀︑︓︄︑︘ is the parent trunk arteriole diameter, W branch. Successive calculations from the outer peripheral retina towards the ONH yielded a single value of the width of the CRA, the Central Retinal Artery Equivalent (CRAE). is mathematical relationship compared remarkably beer than both the sum of the widths and the sum of the squares of the widths of all arteries entering the retina, that since that time had been used as a measure of the general calibre of these vessels. Aer corroborating the original Parr formula, Hubbard et al. () adapted this approach to deliver an analogous formula for venules that calculated the width (in μm) of a parent vein from the widths of its two branches:
Overview of principal literature
ﺾ ﺾ (.) (ﻁﺼ.ﺼﻁﻀ + ﺾWﺽﻅ.ﺼ + ﺽWﺾﻃ.ﺼ)W︕︄︈︍ = ｮ the ﺾthe narrower and W ﺽwhere, respectively, W︕︄︈︍ is the parent trunk venule diameter, W wider branch. Likewise, the general venular calibre is summarised in a value termed the Central Retinal Vein Equivalent (CRVE). Hence, the Arterio-Venous Ratio (AVR) could be obtained as per following equation:
CRAE AVR = (.) CRVE AVR is dimensionless, thus has advantages over absolute vessel width measurements. Apart from the fact that - by definition - it represents a generalised vessel calibre, rather than isolated vessel diameters, it additionally does not require to take into consideration any scaling differences between different refractive errors of eyes, as these are cancelled out. Correction for refraction is surely important for quantifying absolute retinal vessel widths without errors, but this is not the case when using the AVR (Wong et al., ; Paon et al., ).
Hubbard and his colleagues modified Parr’s methodology in a way to make it more aractive to use in large population studies, like the Atherosclerosis Risk In Communities (ARIC) study (Hubbard et al., ) in which they first tested its validity and reproducibility. ey first defined a ring-shaped measurement zone, concentric to the optic disk and half to one Disk Diameter (DD) away from it (Figure .). en, instead of identifying the pairs of every branch vessel and the corresponding parent trunk as in the Parr method, they arbitrarily matched the largest vessel with the smallest one, then the next largest with the next smallest and so on, until all vessels coursing through that measurement zone were accounted for and the central retinal equivalents (CRAE, CRVE) were calculated and from these, AVR. In case the number of vessels to be combined is odd, the remaining single vessel is carried on to the next iteration. A comparison between the modified ARIC method with the original Parr method showed no statistically significant differences between AVR values, hence its use was deemed appropriate.
Four years later, Knudtson et al. (), essentially members of the previous ARIC study, suggested revised formulæ, as well as methodology, for summarising retinal vessel diameters, proving their superiority over the Parr-Hubbard formulæ and methodology. e major advantage of the revised formulæ over the previous ones is that they do not contain constant values, thus can be solved for various measurement units (eg. number of pixels) and are not constrained for vessel widths in micrometers only, being virtually independent of image scale. Regarding the methodology, instead of measuring all vessels lying within the measurement zone, they only include the six largest of them in their formulæ. Investigating the relationship
Overview of principal literature
Figure .: Concentric AVR measurement rings as defined by the ARIC study (Hubbard et al., ). e grid is composed of three circles concentric with the ONH: the innermost circumscribing the average optic disc, the middle one including the area from the disc margin to half DD from the margin and the outer one including the area from half DD to DD from the disc margin. between the number of vessels taken into the measurements and the resulting CRAE and CRVE they found a strong increasing trend that falsified the final result. erefore, restricting the measured vessels at six at all times their methodology proved more robust and at the same time reduced the process time of the calculations. Several studies implemented the use of these revised formulæ (Taarnhøj et al., ; Cheung et al., ).
Of course, the various approaches of different investigators are not ceasing to evolve, complic- ating the quest for standardisation. Paon et al. (Paon et al., ) proposed another revised formula that incorporates an asymmetry factor of the retinal arteriolar branching. Various formulæ for calculating AVR have been tested and compared (Hemminki et al., ) and more recently newer methods have emerged as well that incorporate extended measurement zones (Cheung et al., ).
... Vessel Tortuosity Index
In healthy subjects, blood vessels follow a fairly straight course or are only slightly, in an arc-fashion, curved. One can find different definitions of tortuosity indices in the literature (Kalitzeos et al., ). e most easily implemented and widely used, the tortuosity index, T, is calculated as the ratio of the actual length of the vessel segment L (arc length) to the straight line distance between two branching points, D (chord length).
L T = (.) D
Overview of principal literature
e bending of a vessel influences its local flow haemodynamics and may result in adverse clinical consequences. us, quantifying tortuosity could be used as an indicator of retinal morphological changes, either on a local extent if specific vessels are chosen or globally if the total vascular tree is analysed.
Early investigations (Moore, ) on the signs of arteriosclerosis were rather contained in including tortuosity of the arteries as one of them, because of its wide variability under physiological conditions and its rare occurrence. Various theories have been described (Bracher, ) and tube models tested (Kylstra et al., ) to try to elucidate the aetiology of vessel tortuosity. In vascular disease the vessel wall loses its elasticity and the lumen is narrowed, impeding blood flow, as previously described (Section ...). As a result of the force of BP upon a tube which has lost its carrying power, the vessel then becomes tortuous (Bracher, ; Gunn, ). Leatham (Leatham, ) stated that tortuosity of retinal arterioles was not believed to be a sign of hypertension because it occurred randomly in both aged, normotensives individuals as well as hypertensives. Adding to the uncertainty of accessing general vessel tortuosity during ophthalmoscopy, the subjective characterisation such as “not noticeably tortuous”, “moderately tortuous” and “markedly tortuous” was making its usefulness even more dubious.
Later investigators incorporated manipulation of fundus photographs with the use of bulky, but quite accurate devices such as profile projectors (Lotmar et al., ). Absolute and relative measurement of retinal arterial tortuosity was made possible by subdividing a vessel into a series of circular arcs of individual curvature and measuring the chord lengths and arrow height of these arcs. e sequence of chords is considered to represent the vessel in its “non-twisted” form. It is easy to understand that such a technique was entirely manual to perform, thus time-consuming, error-prone and difficult to reproduce. Others (Kylstra et al., ) defined tortuosity as merely the sum of the height of the arcs that make up the tortuous vessel.
e course of arterioles from projected fundus images on a digitising table was measured from two normotensive age groups (Williams, ). Measurement of a single arteriole (∼mm long), representative of all arterioles in the posterior pole region was performed by calculating the distance of its actual path length and the distance of the line connecting its first and last point. e tortuosity index is the ratio between them as previously mentioned (Equation (.)). An absolutely straight vessel would have an index of , whereas one that has an actual course % longer than its straight line course would have a tortuosity of .. Differences between old and young age groups were not significant and other investigators confirmed the same outcome (Taarnhøj et al., ), although their methodology differed significantly.
Overview of principal literature
e above definition of the tortuosity index has been described as unsuitable for its weakness to distinguish between vessels with equal path lengths and different degree of bending, so a number of novel indices have been proposed (Azegrouz et al., ; Dougherty and Varro, ; Hart et al., ). Advances in computer-assisted methods (Wallace, ; Cheung et al., ; Dougherty et al., ) have pushed current methodologies forward and qualitative studies are currently an exception to the rule (Taarnhøj et al., ).
... Bifurcation Angles and Junction Exponents
A vessel branching can be described geometrically in terms of its junction exponent, x, and its bifurcation angle ω, which is the angle between two daughter vessels. e junction exponent vessels through ﺾd ,ﺽand daughter d ﺼprovides an index of the relative widths of the parent d the equation: ︗ ︗ ︗ (.) ﺾd + ﺽd = ﺼd the narrower one. It has been suggested that, in an ﺾis the wider branch and d ﺽwhere d “ideal” vascular network, there is an optimal way that vessels at bifurcation junctions can be interconnected, in order for fastest transport of blood to be achieved for the least amount of biological work (Murray, ). In other words, it is implied that deviations from optimal vascular architecture may be associated with vascular damage. ese theoretical, optimal :so Equation (.) is wrien ,°ﻁﻃ ⋍and ̂ω ﺿ = values were calculated to be x
ﺿ ﺿ ﺿ (.) ﺾd + ﺽd = ﺼd
Assumptions made when deriving Equation (.) include laminar blood flow and constant blood viscosity. Comparison of experimental results to theoretical values have followed, although in scarce numbers. Zamir and colleagues (Zamir et al., ) introduced two non- dimensional parameters in order to make comparisons of relative diameters of branches rather than absolute diameters: the area ratio β and the asymmetry ratio α; in that way differences of magnification between images was not an issue. Even so, results followed the trend of the theoretical values, but scaered significantly.
Few studies have considered comparing healthy subjects with hypertensives. Average bifurc- ation angles from both normotensives and hypertensives varied considerably from theoretical values (Stanton et al., ), but at the same time the sample size was small and only arteries were measured. Bifurcation angle values declined with increasing age. Regarding junction exponents, both groups had similar values, declined with age and were always smaller than the theoretical value of . Another study’s results (Houben et al., ), followed an almost identical trend for arterial bifurcation angles between hypertensives and normals showing significant
Overview of principal literature differences. eir aempt to measure vein bifurcation angles as well, yielded insignificant differences.
In a rather small sample, the effect of oxygen and carbon dioxide inhalation was tested in normotensives and hypertensives (Chapman et al., ). Neither bifurcation angles, nor junction exponents differed significantly between the two BP groups and parent arteriolar diameters were comparable. Hypertensive patients had a less marked arteriolar constriction when breathing oxygen than controls and breathing carbon dioxide resulted in increased arteriolar diameters in healthy subjects, but not in the hypertensive group. No alterations in junction exponents in either groups were noted.
Comparison of branching angles between individuals with atherosclerosis and healthy subjects showed no significant difference (Chapman et al., ). Instead of using junction exponents, the same group introduced a novel optimality parameter that performed beer in terms of reliability and showed significant difference between the two groups. Recently, another study (Wi et al., ) proposed a parameter referred to as optimality ratio, that appears promising.
... Length-to-Diameter Ratio
e Length-to-Diameter Ratio (LDR) is another structural, dimensionless quantity that may describe the retinal vascular bed. As the name suggests, it is defined as the ratio of the length of a vessel, between two branching points, to its diameter over that segment (King et al., ). Its clinical usefulness has not been extensively assessed as yet (Chapman et al., ; Hughes et al., ).
.. Assessing Retinal Vessel Dynamics
All structural parameters described in the previous sections are extracted from measurements based on static fundus images which essentially are a snapshot of the constantly changing retinal circulation. e opportunity to monitor microvasculature over a period of time might elucidate more complex functional principles.
... Retinal Vessel Analyser
e Retinal Vessel Analyser (RVA) is a commercially available (Imedos Systems, Germany) ophthalmoscopic instrument capable of acquiring vessel diameter fluctuations in real time. ese diameter changes happen physiologically due to the pulsatile nature of blood flow and additionally may be altered by means of external provocation. RVA comprises a mydriatic fundus camera, a Charged-Coupled Device (CCD) video camera and a personal computer that uses dedicated soware to control and adjust the measurement parameters. Versatility is a
Overview of principal literature key aspect of RVA, being easily extendible with a flicker module for vascular reactivity testing (Section ...), an oxygen module for Oxygen Saturation (OSat) mapping (Section ..), an electrocardiograph or a BP monitoring interface for synchronisation with the cardiac cycle (Blum et al., ) and with a video recorder for offline post-processing of the recorded sessions.
︔︒ mydriatic fundus camera capable of acquiring︋️ﺼﻁﻀAt the system’s heart is a Carl Zeiss FF both full color and red-free (- nm) retinal images °, ° or ° wide. Still images can be recorded at a resolution of × pixels, while video sequences are displayed at a lower resolution of × pixels. To achieve an optimum contrast for visualisation of the retinal blood vessels a special green filter is intercepted in the illumination pathway of the fundus camera. us, green light enters the subject’s eye via its pharmacologically dilated pupil. Since, retinal blood vessels containing haemoglobin have different absorption spectra from the surrounding tissue, the integrated vessel tracker registers with the red blood cells column and follows it throughout the course of time making both temporal and spatial diameter analysis possible. Consequently, a data matrix of vessels’ diameters is obtained at the end of a measuring session. Temporal resolution of the RVA system is msec (i.e. diameter readings per second), while spatially it assesses one mean diameter value every Measuring Units (MU); where one MU corresponds to one micron for the standard Gullstrand eye and assuming relaxed accommodation. Calculation of relative values, for example baseline versus stimulation values, minimises the influence of deviation of individual eyes from the Gullstrand eye model, as well as from optical errors.
Although subject compliance is crucial for taking quality measurements, small eye and/or head movements, as well as transient shadows or reflections are unavoidable. To overcome these practical issues, the RVA encompasses adaptive algorithms that can compensate for a reasonable amount of such disturbances (Muench et al., ). e view obtained from the fundus camera is simultaneously acquired from the -CCD video camera (JVC, KY-FBU), displayed on the computer’s monitor and optionally recorded on video tape at the same time.
Regarding continuous baseline vessel diameter recordings, short term and day to day repro- ducibility of the RVA system have been found to be higher for veins than arteries for a five minute long measurement session (Polak et al., ). Nevertheless, short term reproducibility was reported to be excellent with Intraclass Correlation Coefficient (ICC) values of . for veins and . for arteries and slightly lower for day to day sessions (. and ., respectively). Similarly high reproducibility values for baseline (i.e. constant illumination) diameter measurements have been reported a few years later as well (Pache et al., ).
Overview of principal literature
... Flier Provocation
Contrary to extraocular blood vessels, intraocular retinal vessels are not innervated (Brown and Jampol, ), thus depend on local autoregulatory factors for actively regulating blood flow (see Section ..). e autoregulation mechanism, that maintains constant blood flow despite changes in arterial perfusion pressure or metabolic demands of the surrounding tissue, is an undoubtful and well documented phenomenon (Johnson, ). Retinal vessels are capable of responding to such changes by either vasodilation or vasoconstriction, accordingly (see Section ..). Despite the complex association between endothelial function and Cardio- Vascular Disease (CVD) progress (Luscher, ), impaired retinal vascular reactivity has been described in a number of pathological conditions; hypertension amongst them (Panza et al., ; Delles et al., ).
Following animal studies, Formaz and collaborators (Formaz et al., ) proved an increase in retinal vessel diameters induced by diffuse luminance flicker illumination in the human retina. Since then, many studies have exploited retinal flicker provocation as a tool to assess vascular reactivity in health and disease (Heitmar and Summers, ). Interrupting the illumination path of the RVA system with an optoelectronic shuer, light is modulated with a rectangular bright-dark (on/o) wave to produce a flicker stimulus over the entire ° angle field of the camera. is flicker stimulus has a . Hz frequency, well within the range of frequencies (- Hz) where the human visual system’s sensitivity for luminance flicker is at its maximum (Lee et al., ). For a PAL standard CCD video camera of a frame rate of Hz, every other frame is a dark image, thus halving the amount of collected data points during the flicker cycles.
Across literature there are reports of various flicker protocols being implemented. Prior to the adaptation of the embedded RVA flicker module, research groups used their own implementations of external flicker stimuli (Polak et al., ). ese used not only different flicker frequencies ( Hz), but also different baseline/flicker durations and amount of flicker repeats. Since , the standard and most widely used flicker protocol is the one shown in Figure ., using the . Hz flicker frequency. Starting off with a baseline of seconds, the first flicker period of seconds starts. e paern of seconds of recovery (i.e. baseline illumination) and seconds of flicker is then repeated two times and the measurement concludes at the seconds mark, or put differently, in five minutes and fiy seconds.
baseline flicker recovery flicker recovery flicker recovery 0 50 70 150 170 250 270 350 (seconds)
Figure .: Schematic representation of the standard flicker measurement protocol.
Overview of principal literature
For practical reasons of avoiding artificially variable baseline vessel diameters, it is common practice to discard the first seconds of every measurement session from subsequent analysis.
A typical response of a pair of retinal vessels to the RVA flicker session is shown in Figure .. Arteries and veins follow distinct reaction paths, owing to the functional and structural differences between each vessel type. For the artery, with each flicker initiation the vessel responds with vasodilatition, relying on the principles of neurovascular coupling (Riva et al., ). Aer flicker cessation, artery diameter decreases below baseline values (reactive constriction) and returns to the range of initial diameter values approximately - seconds post flicker cessation. Essentially, the choice of seconds flicker duration and seconds of recovery time thereaer is empirically based on a compromise between gaining a marked vessel response and allowing enough time to vessels for their diameters to return back to initial values (Nagel et al., ). Veins, on the other hand, do not undergo a constriction phase as arteries do and additionally show a sustained vasodilation phase, characteristic of their larger vessel compliance and their nature of being reservoir vessels.
... Dynamic Response Analysis
Analysis of the retinal arteriolar and venular responses similar to the one in Figure . can be performed either by utilising the RVA soware-generated output or by independently processing raw data. e RVA soware summarises the three flicker provocations into one average and reports the following four parameters: A︌︀︗, A︌︈︍, A️︄︀︊ (maximum arterial dilation, constriction and difference between the two, respectively) and V︌︀︗ (maximum venular dilation). For statistical purposes, these responses are calculated from an arbitrarily chosen time window, that encompasses seconds: seconds before flicker cessation and seconds aer (Kotliar et al., ). en, the soware automatically generates a report, classifying the vessel responses according to values from Table ., as “normal”, “narrowed”, “unremarkable” or “undetectable” reaction.
Arteries Normal Mean Normal Standard Deviation (SD)
A︌︀︗ (%) . ±. A︌︈︍ (%) -. ±. A️︄︀︊ (%) . ± Veins
V︌︀︗ (%) . ±.
Table .: Classification scheme of vessel response to flicker used by the in-built RVA soware. Based on these values (Nagel et al., ), RVA generates a report of a “normal”, “narrowed”, “unremarkable” or “undetectable” reaction.
Overview of principal literature
103 artery 102 vein
101
100
99
98
97 Diameter (MU)
96 flicker flicker flicker
95
94
93 50 70 150 170 250 270 Time (sec)
Figure .: A typical, normal arterial (red) and venular (blue) response from flicker provocation following the standard protocol shown in Figure .. Doed vertical lines indicate the presence of the flicker stimulus. e first seconds have been discarded prior to ploing the graphs.
In terms of analysing raw data output independently of the RVA soware, there are various approaches as well as measurement parameters calculated. e basic distinction between approaches is choosing to report on an averaged response across all three (or as many as possible) flicker repeats or choosing to analyse them separately. Irrespectively of that choice, the parameters used to characterise an arteriolar and venular response to flicker provocation are the following:
• Baseline Diameter Fluctuation (BDF), the maximum amplitude (peak-to-peak) for arteri- oles and venules, seconds prior to (each) flicker start
• Maximum Dilation (MD), the maximum -second diameter for arterioles and venules, within seconds aer (each) flicker start
• Maximum Constriction (MC), the minimum -second diameter for arterioles, within seconds aer (each) flicker start
• Dilation Amplitude (DA), the difference between MD and MC for arterioles
• Baseline-Corrected Flicker Response (bFR), the difference between DA and BDF for arterioles (Nagel et al., )
• Reaction Time (RT), the time needed (in seconds) to reach MD for arterioles and venules
• Constriction Time (CT), the time needed (in seconds) to reach MC for arterioles
• ΔD, the difference between MD and the -second vessel diameter prior to flicker start for arterioles or venules (a measure of vessel dilatory capacity) (Heitmar et al., )
Overview of principal literature
• Average Peak Ratio (APR), the ratio of DA over BDF for arterioles (a measure of vessel elasticity) (Heitmar et al., a)
All diameter values are reported as a % change (i.e. normalised) to the initial (i.e. prior to the first flicker) mean baseline diameter. Similarly to the MD and MC parameters, others have defined a seconds time window (calculating the median), i.e. seconds before and seconds aer the time point of the maximum/minimum dilation/constriction diameter and have termed it “mean maximal dilation/constriction”, respectively (Kotliar et al., b). Also, the area under the reaction curve during baseline and during/aer flicker has been calculated, essentially providing an average diameter (Gugleta et al., ).
... Other External Provocations
e first report on the results of external provocation with the RVA system by means of isometric exercise was performed in (Blum et al., ) on healthy volunteers with encouraging results for the feasibility of retinal functional assessment. Other techniques that temporarily alter retinal blood flow to assess vascular reactivity include the use of an oculo- oscillo dynamograph (Nagel and Vilser, ; Kotliar et al., ) and gas mixtures inhalation (Blum et al., ; Kiss et al., ; Resch et al., ; Wimpissinger et al., ; Jean-Louis et al., ). A combination of stimulation techniques is also possible; performing an isometric exercise (Bek et al., ; Jensen et al., ) or temporarily elevating IOP by means of an episcleral suction cup (Garhöfer et al., ) and at the same time stimulating the retina with flickering light.
.. Retinal Oximetry
Measurement of retinal oxygen consumption may provide important clinical information about the metabolic state of the retina. Differences between the oxygen delivered to the retina via arterioles and drained away from it via venules can be quantified, if the total OSat in these vessels is measured. Such information can be used to complement our understanding on retinal function in health and disease.
Dual-wavelength oximetry uses digitally recorded retinal images obtained simultaneously at two distinct wavelengths to determine retinal vessel OSat. A dual band-pass filter at a sensitive ( nm) and a non-sensitive ( nm, isosbestic) wavelength is replacing the optoelectronic shuer in the previously described RVA camera system (Section ...). Oxyhemoglobin and deoxyhemoglobin (Hb) absorb light equally at the isosbestic wavelength, whereas (ﺾHbO) there is a considerable absorbance variation at the sensitive wavelength. Since it is practically impossible to measure neither transmiance nor absorption of light in the retina, in order
Overview of principal literature to determine OSat, reflection must be exploited instead. us, Optical Density (OD) can be calculated based on the retinal reflectance of the retinal vessels (I︈︍︓) and that of the surrounding tissue (I︄︗︓), measured as grayscale pixel values:
I OD = log ︄︗︓ (.) I︈︍︓
e Optical Density Ratio (ODR) is the ratio between the ODs of the two sampled wavelengths, which has been found to be linearly related to OSat aer compensation for vessel diameter and fundus pigmentation (Hammer et al., ). Reliability and reproducibility of the technique has been recently evaluated in healthy volunteers with promising results (Lasta et al., ; Man et al., ).
.. Visual Field Testing
Visual Field (VF) testing (or else perimetry) is an additional ancillary test for following func- tional changes of many retinal diseases, alongside its frequent use in glaucoma management and in diagnosis of neurological disorders. Static automated perimetry is the most common method of clinical VF testing. It involves determining the dimmest stimulus that can be seen at a series of pre-determined test point locations. e Humphrey Field Analyser II (Carl Zeiss Meditec, Germany) is an advanced, commercially available perimeter intended to identify visual field defects. It incorporates widely accepted testing algorithms, especially the Swedish Interactive resholding Algorithm (SITA), which offer very high accuracy and a relatively short test time. e test of choice for examining the central visual field is the - SITA Standard paern which comprises test point locations covering the central ° field with a grid of points ° apart. Two quantitative global indices that summarise visual field status are the mean deviation and the paern standard deviation, both measured in decibels. e former shows how much, on average, the whole field deviates from normal while the laer reflects irregularities in the field, such as those caused by localised defects. ese can be used in research to sort eyes into groups of varying disease states.
. Nailfold Capillaroscopy
Structural evaluation of the morphology, distribution and number of capillaries is deemed necessary to investigate microvessel rarefaction at a peripheral level. Capillary rarefaction occurs in many tissues in patients with essential HT and has been shown to contribute to an increased peripheral vascular resistance (Serne et al., ). It is yet unknown whether abnormalities in these vessels are a cause or consequence of elevated BP (Noon et al., ). Rarefaction may be caused by a structural (anatomic) lack of capillaries, functional non-
Overview of principal literature perfusion, or both. Conjuctival (Harper et al., ) and nailfold (Noon et al., ; Antonios et al., ) capillary density have been found significantly smaller in hypertensives than in normotensives. Common techniques used to assess cutaneous microvascular function include capillaroscopy, venous occlusion plethysmography, and laser Doppler anemometry (Yvonne- Tee et al., ).
e nailfold plexus is one of only a few locations on the human body where capillaries advance close enough to the skin surface to become easily detectable in vivo. ey lie in hairpin-like loops parallel to the skin surface. Each loop consist of an arterial and a venous limb. Using laser Doppler anemometry blood cell velocities can be measured within single skin capillaries, not only from the nailfold area, but also in the sublingual and lip skin. Of course, the nailfold area causes no discomfort to the patient and is easier to minimise involuntary movements, thus it is the preferred microvascular sampling site.
.. Measurement Principle
When a narrow beam of laser light is focused onto an arterial or venous limb of a capillary loop, a fraction of the laser light is backscaered by red blood cells, shiing the frequency of light according to the Doppler effect. e frequency shi is directly proportional to the speed of the blood column. Impaired haemodynamic paerns and functional activity of the nailfold capillaries have been reported aer topical temporary cooling of the area. e test is and directing it to the fingertip by means of a ﺾbeing performed by rapidly decompressing CO tube. e nailfold capillaries are being observed and blood velocity is recorded before, during and aer the cooling phase. Flow stop is defined as velocity below . mm/sec for longer than seconds (Mahler et al., ). Prolonged stop of flow, a condition termed vasospasm, characterised hypertensives contrary to normotensives in a study (Gasser and Bühler, ). is finding has been interpreted as a reduced functional reserve at a capillary level in HT.
.. CapiScope Capillaroscopy System
e CAM Laser Doppler Capillary Anemometer (KK Technology, England) is a commercially available video capillaroscopy system that can measure blood cell velocity (Stücker et al., ). It uses a low power, near-infrared ( nm) laser for detecting the necessary Doppler shi. e use of this wavelength has several advantages: it exhibits a deeper penetration depth, the measurement is less affected by the oxygenation of blood and it is also less dependent on skin color, as melanin has a low absorption in the near-infrared. Actual sample depth is typically μm. e laser beam is focused via the objective down to a microns diameterﺼﺼﺽ less than spot in the centre of the field of view. Illumination is provided by LEDs emiing green light ( nm) to maximize the contrast between the erythrocytes and the surrounding tissue. A
Overview of principal literature
CCD camera (Model XC-CE, Sony, Japan) needs to be focused so that the object plane and the laser focal point match to get a clear view. e camera output is fed to a computer monitor.
e CAM system provides an approximately × magnified image of the nailfold plexus with a resolution of × pixels. e device can be positioned appropriately onto the limb of a capillary loop using an XYZ micropositioner stage. e laser beam is reflected by blood cells at the focal point moving parallel to the tissue surface. is gives the laser light a Doppler shi directly proportional to the velocity of the reflecting blood cells. Acoustic feedback control with sound from the Doppler shi provides real time audible cues via the computer speakers during the entire measurement. is allows the operator to obtain the point of maximum signal strength more readily. e Doppler shied laser signal is collected by the objective and internal optoelectronics and processed in real time by means of the accompanied CapiScope soware producing a velocity trace (in absolute units of mm/s). is system can be used to measure velocities ranging from . to . mm/s. e velocity trace can be saved along with an image or video sequence of the capillary being measured, onto the hard drive of the computer, for offline post-processing.
Chapter
Retinal Vessel Analyser: Reproducibility
. Baground
During the early years of introduction of the RVA (Chapter ) into the scientific seing, it was merely used to observe and record retinal vessel diameters under constant illumination (baseline conditions) over time. A few studies, as mentioned in Section ..., have reported on the device’s reproducibility and sensitivity under such conditions with excellent results. How- ever, with the introduction of flicker stimulation and the possibility to probe retinal reactivity, a completely different regime of measurements (dynamic reaction) was characterising RVA’s main function. Unfortunately, subsequent studies utilising flicker measurements described the accuracy of the system by referring to the reproducibility and sensitivity of the non-relevant early studies.
Studies actually reporting on reproducibility of the flicker responses are either irrelevant to the current hardware and protocol or incomplete. An early study reported Coefficient of Variation (CV) values of % but it was unclear whether this was for arteries, veins, or combined (Polak et al., ). A later study reported CV values of .% for arteries and .% for veins (Garhöfer et al., ). However, both studies at that time were using a prototype RVA system and a flicker frequency of Hz. Nagel et al. (a) were the first to report short- term ( hour) and long-term ( month) variability of flicker responses but had the following limitations: they measured only one parameter (maximum vessel dilation), they did not state how they defined that parameter and also they averaged responses over the three flicker cycles within each session. Reproducibility of the values calculated with the inbuilt RVA analysis have been reported in healthy Asian individuals (Nguyen et al., ) using non-standard measures (Pearson correlation coefficients). e only report - using the current RVA system and the current protocol - on CV values among both baseline and reaction diameters has been published
Retinal Vessel Analyser: Reproducibility only recently, with low CV values for MD and MC and moderate CV values for RT (Heitmar et al., ).
.. Motivation and Resear Rationale
e RVA in its current implementation is a fairly new research tool in retinal functional assessment since it was introduced no earlier than (Nagel et al., a). A few years later, several experts in the field published a feature review on the RVA and its applicability, highlighting several “unresolved open questions” (Garhöfer et al., ). e insufficiency of reproducibility data of flicker responses was the main one.
e nature of such measurements, being related to microvasculature haemodynamics, is inherently variable. us, protocols should be strictly adhered to and standardisation pro- cedures should always be meticulously observed. Otherwise, external factors might mask or exaggerate the true flicker responses. On that basis, detailed reproducibility analysis by means of the ICC is performed: comparisons between examiners are reported for the first time and the intraobserver (or else intersession) reproducibility analysis from Heitmar et al. () is expanded upon with a greater amount of parameters tested across arteries and veins on a flicker per flicker breakdown, as well as on averaged flicker cycles.
Lastly, the vendor-generated flicker-reaction report classifies responses according to an ob- solete flicker protocol (Nagel et al., ) and its inbuilt parameters are defined on a rather narrow time window of seconds ( seconds prior to flicker cessation and seconds aer) (Section ...). Recent studies utilising different parameter definitions (Heitmar et al., ) have reported values of maximum dilation and/or constriction to be occurring outside this time frame, thus rendering the appropriateness of the RVA-generated parameters ambivalent. us, comparison between the inbuilt and independent analysis output is being performed.
.. Aims
e aims of this study were the following:
• to quantify and test the interobserver and intraobserver reproducibility of the RVA system for the independently analysed dynamic retinal vessel reactivity parameters (BDF, DA, MD, MC, bFR, ΔD, APR, RT, CT) on a per flicker analysis. • to quantify and test the reproducibility of the inbuilt soware-generated parameters of
A︌︀︗, A︌︈︍, A️︄︀︊ and V︌︀︗ and to compare them with their counterparts: arterial MD, MC, DA and venular MD, respectively.
Retinal Vessel Analyser: Reproducibility
• to quantify the static retinal vessel parameters of tortuosity, branching angles and AVR as processed from fundus photos obtained with the same RVA system.
. Subjects and Methods
.. Interobserver Reproducibility
Measurements were performed by two Examiners for healthy volunteers ( males, females) under identical conditions. Both Examiners had comparable experience with the RVA device and adhered to the detailed protocol described in Section ... e sequence of data collection between Examiner and Examiner was arbitrarily selected. A break of at least five minutes was allowed between the two sessions per subject. All measurements took place within months.
.. Intraobserver Reproducibility
Measurements were performed by a single examiner for healthy volunteers ( males, females) on two occasions under identical conditions. All measurements adhered to the detailed protocol described in Section ...
.. Standardisations Applied
e vessel length selection was governed by each individual’s angioarchitecture, but was always kept as long as possible. Nevertheless, when selecting vessel segments for a meas- urement in real time, no two pairs can be selected to be precisely equally long. us, in order to standardise comparisons and to eliminate potential influence of the segment length measured between examiners (interobserver reproducibility study) as well as between sessions (intraobserver reproducibility study) the following procedure was followed: for every pair of vessel segments that underwent comparison, the longer one was truncated to exactly the same length as the shorter one. is was possible by processing the raw data matrix output of the RVA. Moreover, for the repeated measurement throughout the intraobserver reproducibility study, the repetition feature of the soware was used, which automatically measures exactly the same location as in the previous one. In some cases this was not possible due to registration issues, but the location was manually matched as closely as possible.
.. Inclusion and Exclusion Criteria
All participants conformed with the following criteria:
• aged at least years old
Retinal Vessel Analyser: Reproducibility
• had no history of systemic disease or any current relevant manifestation
• were medication-free
• had clear optical media
• had no history of epilepsy
ey were given a minimum of hours to decide on their participation and to ask any questions, aer receiving a wrien description of the study protocol. e tenets of the Declaration of Helsinki were observed, and institutional review board approval was granted. Wrien informed consent was obtained from every participant prior to the measurements.
.. Study Protocol
At least hours prior to their morning visits, participants were asked to abstain from smoking, from consuming products containing alcohol or caffeine, as well as from taking up any sort of considerable physical activity, whereas they were instructed not to fast. Room temperature was maintained constant during all measurements (- ℃).
... Intraocular Pressure Measurement
Non-contact tonometry was performed to assess IOP by means of a validated (Ogbuehi and Almubrad, ) device (Pulsair EasyEye, Keeler Ltd., UK). ree readings were obtained from each eye and the average value was recorded.
... Blood Pressure and Pulse Assessment
All participants remained seated for at least minutes to ensure stable haemodynamic conditions prior to the start of the examination. BP was measured from the brachial artery of the forearm using a validated (Rogoza et al., ), automated oscillometric digital BP monitor (UA-, A&D Instruments Ltd., UK). ree consecutive readings of SBP, DBP and HR were recorded. MABP was calculated as previously defined (Equation (.)).
... Retinal Vessel Functional Assessment
Details on the measuring principle of the RVA have been described earlier (Section ...). One arbitrarily selected eye was examined. One drop of Tropicamide (% w/v, Bausch & Lomb, UK) was instilled to achieve pupil dilation necessary for geing unobstructed view of the posterior pole. As soon as full pupil dilation was reached, the dynamic retinal vessel assessment commenced. e fellow eye was covered to achieve good fixation. During the examination, subjects were encouraged to blink normally (to maintain a sufficiently wet cornea) and to
Retinal Vessel Analyser: Reproducibility maintain steady fixation at the internal target, the tip of a needle. is was placed accordingly in order to position the desired vessel segments centrally as viewed on the computer monitor. e measurement location selected was - DD away from the ONH (Garhöfer et al., ) (Figure .). First, the arterial vessel segment was selected, then the venular one and as soon as the soware registered successfully the corresponding positions the measurement session started automatically. In case the contrast between the vessels and the background tissue was not adequate, or other degradations affecting video quality appeared, the measurement was aborted and restarted. e standard flicker protocol of seconds baseline, seconds flicker and seconds of recovery was applied (Figure .). e soware used throughout the data capturing sessions was the vendor-supplied Retinal Vessel Analyser (V...).
... Retinal Vessel Structural Assessment
Following the RVA assessment, the camera was set to take monochromatic and color fundus images from the same, previously dilated eye. e vendor-supplied soware - Visualis (V...) - was used for capturing the fundus photos. A series of images were obtained as follows:
• for AVR calculation: one monochromatic image with the ONH centered at ° and one color RGB image (for reference)
• for tortuosity and branching angles: one monochromatic image with the macula centered at ° that included the superotemporal and inferotemporal quadrants
e desired topology was achieved by instructing the subject to follow a blinking red fixation Light-emiing Diode (LED) light with the fellow, unobstructed eye. e images were digitally stored in Tagged Image File Format (TIFF) format (lossless) for subsequent analysis.
Figure .: Typical measurement location. Leer A in the red circle represents the arterial vessel segment and leer V in the blue circle represents the venular vessel segment.
Retinal Vessel Analyser: Reproducibility
For the AVR measurement, the vendor-supplied soware was used for analysis, namely VesselMap (V...). e CRAE and the CRVE were semi-automatically obtained from each subject according to the ARIC protocol (Hubbard et al., ). e quotient of these values was automatically calculated to derive AVR. At least four arteries and four veins were selected for the AVR measurement, depending on individual angioarchitecture. Whenever images were captured on two occasions, average AVR values were calculated (Figure .).
For the measurement of branching angles (Figure .) and tortuosity values (Figure .) ImageJ (Version .v) (Abramoff et al., ) was used on monochromatic fundus images for increased contrast and, hence, greater accuracy. First and second order major bifurcations of one representative artery and one representative vein were selected distal to the ONH for branching angles and the vessel segment linking those two bifurcations defined the location of tortuosity measurements. e line coursing through the vessels measured the arc length, whereas the straight line corresponded to the chord length. e ratio of the two equalled the tortuosity index. All bifurcation angle and arc/chord length measurements were repeated three times and their averages went into subsequent analyses.
.. Outcome Measures
.. Dynamic Parameters
e following parameters (as defined in Section ...) were measured and compared (between examiners and between sessions, for inter- and intraobserver reproducibility respectively) for all three flicker cycles (henceforth designated with numbers , and ) by means of raw RVA output data processing: BDF, bFR, MD, MC, DA, RT, CT, ΔD and APR. Absolute arteriolar and venular diameters were recorded in MU and compared across Examiners and across measurement sessions. Also, the RVA generated parameters of A︌︀︗, A︌︈︍, A️︄︀︊ and
Figure .: Measurement of AVR of a right eye by manually selecting arteries (red) and veins (blue) coursing through the outer ring (le). Measurement of AVR of the same eye from a subsequent visit (right). See also Figure . on page .
Retinal Vessel Analyser: Reproducibility
Figure .: Illustration of bifurcation angles measurement using ImageJ. Values reported are in degrees. e angle marked in green represents the proximal measurement site and the angle marked in yellow represents the distal measurement site. e two bifurcation angles define the vessel segment taken for tortuosity measurements (see Figure .).
Figure .: Illustration of tortuosity measurement using ImageJ, of the same vesssel segment as in Figure .. Values reported are dimensionless. e yellow line marks the arc length and the green line marks the chord length.
Retinal Vessel Analyser: Reproducibility
V︌︀︗ are reported and compared to their counterparts: arterial MD, MC, DA and venular MD, respectively.
.. Static Parameters
CRAE, CRVE and corresponding AVR values, tortuosity indices (see Equation (.)) and bifurcation angles for both arteries and veins are reported according to the previously described protocols (Section ...) for the intraobserver cohort (n=) only (Table .).
.. Statistics and Data Analysis
SPSS (Version . Chicago, SPSS Inc.) was used for statistical analysis and Graphpad Prism (Version .) for ploing purposes. Normality tests were performed on all continuous data by means of the Shapiro-Wilk test, to determine distribution. In case of normal distributions, data are expressed as means (SD) and groups are compared by Student’s paired t-tests. Non- normally distributed data are expressed as medians (Inter-artile Range (IQR)), compared by the Mann-Whitney U test. IQR is calculated as the difference between the third and first quartiles. For multiple comparisons across the three flicker cycles, the non-parametric Friedman test was performed. Forward stepwise multiple linear regression analysis was used to test if any of the static or dynamic variables significantly predicted maximum dilation responses. For all calculations, a P value of < . was considered significant. Reproducibility was tested by means of the ICC (Shrout and Fleiss, ). Box-and-whiskers plots shown ︓︇ percentile andﻁ indicate the median and the IQR, whereas whiskers are drawn down to the ︓︇. Points below and above the whiskers are drawn as black filled dots, indicatingﻁﻅ up to the outliers, where applicable.
. Results
.. Interobserver Reproducibility
... Subjects
Characteristics of the participants (n=) for the interobserver reproducibility part of this study .years old (ﻃ±) are shown in Table .. Mean (±SD) age of the healthy volunteers cohort was
... Retinal Vessels’ Absolute Diameters
e retinal arteriolar and venular absolute diameter values followed a normal distribution. Hence, values are expressed as means (SD). No statistically significant difference was found between Examiners of the absolute arterioles and venules diameters that each one selected for
Retinal Vessel Analyser: Reproducibility
Age IOP SBP DBP MABP HR Subject Gender (years) (mmHg) (mmHg) (mmHg) (mmHg) (pulses/min) f m f f m m f f f f f m m
Table .: Demographics and baseline characteristics of subjects (n=) included in the interobserver study. Subjects are sorted by age in ascending order. For acronyms, see page . the assessment of the retinal microvascular reactivity to flicker light. P values are shown in Table ..
... Inbuilt Dynamic Flier Response Analysis
e parameters A︌︀︗, A︌︈︍, A️︄︀︊ for arterioles and V︌︀︗ for venules generated from the RVA soware (averaged across all three flicker cycles) are shown in Table . per Examiner. Since these values followed a normal distribution, a parametric Student’s t-test was performed to check for differences.
... Independent Dynamic Flier Response Analysis
All dynamic response parameters tested for interobserver reproducibility were not normally distributed. us, values shown are medians (IQR). Statistical significance was sought using the Mann-Whitney U test. Box-and-whisker diagrams were ploed for BDF, MD, MC, DA, bFR, ΔD, APR, RT and CT for arteries (Figures . to .) and for BDF, MD, ΔD and RT for veins (Figures . to .) for all three flickers. e difference of MC of arterioles of the second and third flicker cycles between examiners was statistically significant (p=. and p=., respectively). Statistically significant difference was also found between the MD response of arteries during the last (third) flicker (p=.) (Table .).
e time points of maximum dilation and maximum constriction for arteries (RT, CT) and for maximum dilation for veins (RT) are shown in Table .. Comparisons across Examiners
Retinal Vessel Analyser: Reproducibility
Examiner Examiner Examiner Examiner Subject Arteriolar Diameter (MU) Venular Diameter (MU) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Mean (SD) . (.) . (.) . (.) . (.) ICC . . p value . .
Table .: Comparison of absolute arteriolar and venular diameters between Examiners. MU stands for Measurement Units. Student’s paired t-tests were performed. revealed a difference only for the arterial RT during the second flicker cycle (p=.). Across flicker cycles within Examiners, no statistically significant differences were found. Finally, non-parametric comparisons using the Mann Whitney U test between arterial and venular RT revealed differences only for Examiner (flickers and ), with the veins needing longer time to reach maximum dilation compared to their arterial counterparts.
... Comparison Between Inbuilt and Independent Analysis
To compare the inbuilt RVA soware analysis with the one independently calculated from raw data, the three flicker responses as per Table . were averaged. en, the two were
Parameter (%) Examiner Examiner ICC p value
A︌︀︗ . (.) . (.) . . A︌︈︍ -. (.) -. (.) . . A️︄︀︊ . (.) . (.) . . V︌︀︗ . (.) . (.) . .
Table .: Inbuilt RVA dynamic flicker response parameters compared between examiners (n=). Values are expressed as means (SD) % change to baseline diameter. Student’s paired t-tests were performed. For definitions, see Section ... on page .
Retinal Vessel Analyser: Reproducibility
Examiner Examiner p value Parameter ICC Arterioles (across Examiners) BDF (%) . (.) . (.) . . BDF (%) (.) . (.) . . BDF (%) . (.) . (.) . . Friedman test (within Examiner) . . MD (%) . (.) . (.) . . MD (%) . (.) (.) . . MD (%) . (.) . (.) . . Friedman test (within Examiner) . . MC (%) . (.) . (.) . . MC (%) . (.) . (.) -. . MC (%) . (.) . (.) -. . Friedman test (within Examiner) . . DA (%) . () . (.) . . DA (%) . (.) . (.) . . DA (%) . (.) . (.) . . Friedman test (within Examiner) . . bFR (%) . () . () . . bFR (%) . (.) . (.) . . bFR (%) . (.) . (.) . . Friedman test (within Examiner) . . ΔD (%) . (.) . (.) . . ΔD (%) . (.) . (.) . . ΔD (%) . (.) . (.) . . Friedman test (within Examiner) . . APR . (.) . (.) . . APR . () . (.) . . APR . () . () . . Friedman test (within Examiner) . . Venules BDF (%) . (.) (.) . . BDF (%) . (.) . (.) . . BDF (%) . (.) . (.) . . Friedman test (within Examiner) . . MD (%) . (.) . (.) . . MD (%) (.) (.) . . MD (%) (.) (.) . . Friedman test (within Examiner) . . ΔD (%) . (.) . (.) . . ΔD (%) . (.) . (.) . . ΔD (%) . (.) . (.) . . Friedman test (within Examiner) . .
Table .: Independently analysed RVA dynamic flicker response parameters compared between examiners (n=). Values are expressed as medians (IQR). Mann Whitney U tests were performed for across Examiners comparisons and Friedman tests were performed for within Examiner comparisons. Statistical significance is denoted in bold. For acronyms, see page . Retinal Vessel Analyser: Reproducibility e e e g g g n n n 10 10 10 a a a h h h C C C r r r e e e t t t e e e m m m a a a i i i 5 5 5 D D D % % % 2 3 1 F F F D D D B B B 0 0 0 Boxplot 1 E1 E2 Boxplot 2 Boxplot 1 E1 E2 Boxplot 2 Boxplot 1 E1 E2 Boxplot 2 r r r e e e t 100 t 100 t 100 e e e m m m a a a i i i D D D e e e v v v i i i
t 95 t 95 t 95 a a a l l l e e e R R R % % % 1 2 3
C 90 C 90 C 90 M M M
Boxplot 1 E1 E2 Boxplot 2 Boxplot 1 E1 E2 Boxplot 2 Boxplot 1 E1 E2 Boxplot 2
r 115 r 115 r 115 e e e t t t e e e m m m a a a i 110 i 110 i 110 D D D e e e v v v i i i t t t a a a l l l
e 105 e 105 e 105 R R R % % % 1 2 3
D 100 D 100 D 100 M M M
Boxplot 1 E1 E2 Boxplot 2 Boxplot 1 E1 E2 Boxplot 2 Boxplot 1 E1 E2 Boxplot 2 e e e g g g
n 15 n 15 n 15 a a a h h h C C C r r r e e e t t t e 10 e 10 e 10 m m m a a a i i i D D D % % %
1 5 2 5 3 5 A A A D D D
Boxplot 1 E1 E2 Boxplot 2 Boxplot 1 E1 E2 Boxplot 2 Boxplot 1 E1 E2 Boxplot 2 e e e g g g n n n a a a
h 10 h 10 h 10 C C C r r r e e e t t t e e e
m 5 m 5 m 5 a a a i i i D D D % % % 1 2 3
R 0 R 0 R 0 F F F b b b
Boxplot 1 E1 E2 Boxplot 2 Boxplot 1 E1 E2 Boxplot 2 Boxplot 1 E1 E2 Boxplot 2
15 15 15
1 10 2 10 3 10 D D D a a a t t t l l l e e e
D 5 D 5 D 5
0 0 0 Boxplot 1 E1 E2 Boxplot 2 Boxplot 1 E1 E2 Boxplot 2 Boxplot 1 E1 E2 Boxplot 2
8 8 8
6 6 6 1 2 3 R R R
P 4 P 4 P 4 A A A
2 2 2
0 0 0 Boxplot 1 E1 E2 Boxplot 2 Boxplot 1 E1 E2 Boxplot 2 Boxplot 1 E1 E2 Boxplot 2
Figure .: Combined box-and-whisker and “Examiner - Examiner ” comparison scaerplots (joined with straight lines) showing arteriolar diameter fluctuation and flicker responses (n=) across all three flickers. See Table . for numerical values. For acronyms, see page .
Retinal Vessel Analyser: Reproducibility
Examiner Examiner p value Parameter ICC Arterioles (across Examiners) RT (seconds) () () . . RT (seconds) () (.)† . . RT (seconds) () (.)‡ . . Friedman test (within Examiner) . . CT (seconds) () (.) -. . CT (seconds) () () -. . CT (seconds) (.) () . . Friedman test (within Examiner) . . Venules RT (seconds) (.) (.) . . RT (seconds) () (.)† . . RT (seconds) () ()‡ -. . Friedman test (within Examiner) . .
Table .: Independently analysed RVA dynamic response parameters (n=) compared between examiners. Values are expressed as medians (IQR). Mann Whitney U tests were performed for across Examiners and across vessel type (arteries-veins) comparisons and Friedman tests were performed for within Examiner comparisons. † signifies borderline statistically significant difference (p=.) between arteries and veins for flicker . ‡ signifies statistically significant difference (p=.) between arteries and veins for flicker . For acronyms, see page .
RT1 RT2 RT3 ) ) ) 50 50 50 s s s ( ( ( t t t e e e s s s 40 40 40 n n n o o o r r r e e e 30 30 30 k k k c c c i i i l l l f f f e e e 20 20 20 c c c n n n i i i s s s e e e 10 10 10 m m m i i i T T T 0 0 0 Boxplot 1 E1 E2 Boxplot 2 Boxplot 1 E1 E2 Boxplot 2 Boxplot 1 E1 E2 Boxplot 2
CT1 CT2 CT3
) 50 ) 50 ) 50 s s s ( ( ( t t t e e e
s 40 s 40 s 40 n n n o o o r r r
e 30 e 30 e 30 k k k c c c i i i l l l f f f
e 20 e 20 e 20 c c c n n n i i i s s s
e 10 e 10 e 10 m m m i i i T 0 T 0 T 0 Boxplot 1 E1 E2 Boxplot 2 Boxplot 1 E1 E2 Boxplot 2 Boxplot 1 E1 E2 Boxplot 2
Figure .: Combined box-and-whisker and “Examiner - Examiner ” comparison scaerplots (joined with straight lines) showing arteriolar reaction and constriction times (n=) across all three flickers. See Table . for numerical values. For acronyms, see page .
Retinal Vessel Analyser: Reproducibility e e e g g g
n 10 n 10 n 10 a a a h h h C C C r r r e e e t t t e e e m m m a a a
i 5 i 5 i 5 D D D % % % 1 2 3 F F F D D D B B B 0 0 0 Boxplot 1 E1 E2 Boxplot 2 Boxplot 1 E1 E2 Boxplot 2 Boxplot 1 E1 E2 Boxplot 2 r r r e e e t t t e e e
m 110 m 110 m 110 a a a i i i D D D e e e v v v i i i t t t a a a
l 105 l 105 l 105 e e e R R R % % % 1 2 3
D 100 D 100 D 100 M M M
Boxplot 1 E1 E2 Boxplot 2 Boxplot 1 E1 E2 Boxplot 2 Boxplot 1 E1 E2 Boxplot 2
10 10 10 1 2 3 D D D a a a t t t l l l e e e
D 5 D 5 D 5
0 0 0 Boxplot 1 E1 E2 Boxplot 2 Boxplot 1 E1 E2 Boxplot 2 Boxplot 1 E1 E2 Boxplot 2
Figure .: Combined box-and-whisker and “Examiner - Examiner ” comparison scaerplots (joined with straight lines) showing venular diameter fluctuation and flicker responses (n=) across all three flickers. See Table . for numerical values. For acronyms, see page .
RT1 RT2 RT3 ) ) ) 50 50 50 s s s ( ( ( t t t e e e s s s 40 40 40 n n n o o o r r r e e e 30 30 30 k k k c c c i i i l l l f f f e e e 20 20 20 c c c n n n i i i s s s e e e 10 10 10 m m m i i i T T T 0 0 0 Boxplot 1 E1 E2 Boxplot 2 Boxplot 1 E1 E2 Boxplot 2 Boxplot 1 E1 E2 Boxplot 2
Figure .: Combined box-and-whisker and “Examiner - Examiner ” comparison scaerplots (joined with straight lines) showing venular reaction times (n=) across all three flickers. See Table . for numerical values. For acronyms, see page .
Retinal Vessel Analyser: Reproducibility compared using a parametric Student’s t-test and the results are shown in Table .. For Examiner , all arterial parameters calculated by means of the vendor-supplied RVA soware were significantly underestimated, compared to the respective independently analysed ones. e statistical significance of venular maximal dilation response was borderline (p=.). For Examiner , maximal arterial constriction (p<.) and arterial dilation amplitude (p=.) were found to be significantly different between the two.
... Averaged Flier Responses
All previously calculated parameters were tested for reproducibility on a per flicker basis. Since the majority of research groups are performing averaging across the three flicker cycles, we are also presenting reproducibility results of all previously calculated parameters, this time with averaged values, collapsing data across all flicker repetitions (Tables . to .). Averaged values followed a normal distribution, hence parametric tests were performed to check for differences.
.. Intraobserver Reproducibility
... Subjects
Baseline characteristics of the participants (n=) for the intraobserver reproducibility part of (ﻅ±) this study are shown in Table .. Mean (±SD) age of the healthy volunteers cohort was years old.
Parameter (%) Examiner p value Examiner p value A . (.) . (.) ︌︀︗ . . MD (arteries) . (.) . (.) A -. (.) -. (.) ︌︈︍ <. <. MC -. (.) -. (.) A . (.) . (.) ️︄︀︊ <. . DA . (.) . (.) V . (.) . (.) ︌︀︗ . . MD (veins) . () . (.)
Table .: Comparison (n=) between inbuilt (see Table .) and independent flicker analysis (see Table .). Values are expressed as means (SD) % change to baseline diameter. Student’s paired t-tests were performed. Statistical significance is denoted in bold. For definitions, see Section ... on page .
Retinal Vessel Analyser: Reproducibility
Examiner Examiner Parameter ICC p value Arterioles BDF (%) . () . (.) . . MD (%) . (.) . (.) . . MC (%) . (.) . (.) -. . DA (%) . (.) . (.) . . bFR (%) . (.) . (.) . . ΔD (%) . (.) (.) . . APR . (.) . (.) . . Venules BDF (%) . (.) . (.) . . MD (%) . (.) . (.) . . ΔD (%) . (.) . (.) . .
Table .: Independently analysed RVA dynamic flicker response parameters (n=) compared between measure- ment sessions, averaged across all three flicker cycles. Values are expressed as means (SD). Student’s paired t-tests were performed. Statistical significance is denoted in bold. For acronyms, see page .
Examiner Examiner p value Parameter ICC Arterioles (across Examiners) RT (seconds) () () . . CT (seconds) () () . . Venules RT (seconds) () () -. .
Table .: Independently analysed RVA dynamic response parameters (n=) compared between examiners, averaged across all three flicker cycles. Values are expressed as means (SD). Student’s paired t-tests were performed. Statistical significance is denoted in bold. For acronyms, see page .
Parameter Measurement Measurement p value IOP (mmHg) () () . SBP (mmHg) () () . DBP (mmHg) () () . MABP (mmHg) () () . HR (pulses/min) () () .
Table .: Baseline characteristics of subjects (n=) participating in the intraobserver study. Values are expressed as means (SD). Student’s paired t-tests were performed. For acronyms, see page .
Retinal Vessel Analyser: Reproducibility
... Retinal Absolute Diameters
Comparing the absolute arteriolar diameter between measurement sessions revealed no statist- ically significant difference. On the other hand, absolute venular diameters were significantly different. Nevertheless, ICC values for both vessel types showed excellent reproducibility (See Table .).
Vessel Type Measurement Measurement ICC p-value Arteries (MU) () () . . Veins (MU) () () . .
Table .: Comparison of absolute arteriolar and venular diameters (n=) between the two measurement sessions. Values are expressed as means (SD). Student’s paired t-tests were performed. Statistical significance is denoted in bold.
... Inbuilt Dynamic Flier Response Analysis
e parameters A︌︀︗, A︌︈︍, A️︄︀︊ for arterioles and V︌︀︗ for venules generated from the RVA soware (averaged across all three flicker cycles) are shown in Table . for each measurement session. Non-parametric Mann Whitney U tests for significance testing were performed, since values failed to indicate normal distributions.
... Independent Dynamic Flier Response Analysis
Similarly to the interobserver analysis, all dynamic response parameters tested for intraob- server reproducibility were not normally distributed. us, values shown are medians (IQR). Statistical significance was checked using the Mann-Whitney U test. Box-and-whisker diagrams were ploed for BDF, MD, MC, DA, bFR, ΔD, APR, RT and CT for arteries (Figures . to .) and for BDF, MD, ΔD and RT for veins (Figures . to .) for all three flickers. Outliers are shown with black filled dots, where applicable.
Parameter (%) Measurement Measurement ICC p value
A︌︀︗ . (.) . (.) . . A︌︈︍ -. (.) -. (.) . . A️︄︀︊ . (.) . (.) . . V︌︀︗ . (.) . () . .
Table .: Inbuilt RVA dynamic flicker response parameters compared between measurement sessions (n=). Values are expressed as medians (IQR) % change to baseline diameter. Mann Whitney U tests were performed for across measurements comparisons. For definitions, see Section ... on page .
Retinal Vessel Analyser: Reproducibility
Measurement Measurement Parameter ICC p value Arterioles BDF (%) . (.) . (.) . . BDF (%) . (.) (.) . . BDF (%) . (.) . (.) . . Friedman test . . MD (%) . (.) . (.) . . MD (%) . (.) . (.) . . MD (%) . (.) . (.) . . Friedman test . . MC (%) . (.) . (.) . . MC (%) (.) . (.) . . MC (%) . (.) . (.) . . Friedman test . . DA (%) . (.) . (.) . . DA (%) (.) (.) . . DA (%) (.) (.) . . Friedman test . . bFR (%) (.) . () . . bFR (%) . (.) . (.) . . bFR (%) . () . () . . Friedman test . . ΔD (%) . (.) . () . . ΔD (%) . (.) . (.) . . ΔD (%) . (.) . (.) . . Friedman test . . APR . () . (.) . . APR . (.) (.) . . APR . (.) . (.) . . Friedman test . . Venules BDF (%) . (.) . (.) . . BDF (%) . (.) . (.) . . BDF (%) . (.) . (.) . . Friedman test . . MD (%) . (.) . (.) . . MD (%) . (.) . (.) . . MD (%) (.) (.) . . Friedman test . . ΔD (%) . (.) . (.) . . ΔD (%) . (.) . (.) . . ΔD (%) . (.) . (.) . . Friedman test . .
Table .: Independently analysed RVA dynamic flicker response parameters (n=) compared between measurement sessions. Values are expressed as medians (IQR). Mann Whitney U tests were performed for across measurements comparisons and Friedman tests were performed for within measurements comparisons. For acronyms, see page . Retinal Vessel Analyser: Reproducibility
Measurement Measurement Parameter ICC p value Arterioles RT (seconds) ()† ()¶ . . RT (seconds) ()‡ () -. . RT (seconds) ()§ () . . Friedman test . . CT (seconds) () () . . CT (seconds) () () . . CT (seconds) () () -. . Friedman test . . Venules RT (seconds) ()† ()¶ . . RT (seconds) ()‡ () . . RT (seconds) ()§ () . . Friedman test . .
Table .: Independently analysed RVA dynamic response parameters (n=) compared between measurement sessions. Values are expressed as medians (IQR). Mann Whitney U tests were performed for across measurements and across vessel type (arteries-veins) comparisons and Friedman tests were performed for within measurements comparisons. † signifies statistically significant difference (p=.) between arteries and veins for flicker , measurement . ‡ signifies statistically significant difference (p<.) between arteries and veins for flicker , measurement . § signifies statistically significant difference (p=.) between arteries and veins for flicker , measurement . ¶ signifies statistically significant difference (p=.) between arteries and veins for flicker , measurement . For acronyms, see page .
Retinal Vessel Analyser: Reproducibility
25 25 25 e e e g g g n n n a a a 20 20 20 h h h C C C r r r e e e t t t 15 15 15 e e e m m m a a a i i i 10 10 10 D D D % % % 2 3 1
F 5 F 5 F 5 D D D B B B 0 0 0 Boxplot 1 M1 M2 Boxplot 2 Boxplot 1 M1 M2 Boxplot 2 Boxplot 1 M1 M2 Boxplot 2 r r r e e e t t t
e 110 e 110 e 110 m m m a a a i i i D D D e e e v v v i i i t 100 t 100 t 100 a a a l l l e e e R R R % % % 1 2 3
C 90 C 90 C 90 M M M
Boxplot 1 M1 M2 Boxplot 2 Boxplot 1 M1 M2 Boxplot 2 Boxplot 1 M1 M2 Boxplot 2
r 130 r 130 r 130 e e e t t t e e e m m m a a a i i i
D 120 D 120 D 120 e e e v v v i i i t t t a a a l l l e e e
R 110 R 110 R 110 % % % 1 2 3 D D D
M 100 M 100 M 100 Boxplot 1 M1 M2 Boxplot 2 Boxplot 1 M1 M2 Boxplot 2 Boxplot 1 M1 M2 Boxplot 2
e 20 e 20 e 20 g g g n n n a a a h h h
C 15 C 15 C 15 r r r e e e t t t e e e
m 10 m 10 m 10 a a a i i i D D D
% 5 % 5 % 5 1 2 3 A A A D D D 0 0 0 Boxplot 1 M1 M2 Boxplot 2 Boxplot 1 M1 M2 Boxplot 2 Boxplot 1 M1 M2 Boxplot 2
10 10 10 e e e g g g n n n a a a h h h
C 5 C 5 C 5 r r r e e e t t t e e e m m m a a a i i i
D 0 D 0 D 0 % % % 1 2 3 R R R F F F b b b -5 -5 -5 Boxplot 1 M1 M2 Boxplot 2 Boxplot 1 M1 M2 Boxplot 2 Boxplot 1 M1 M2 Boxplot 2
20 20 20
15 15 15 1 2 3 D D D
a 10 a 10 a 10 t t t l l l e e e D D D 5 5 5
0 0 0 Boxplot 1 M1 M2 Boxplot 2 Boxplot 1 M1 M2 Boxplot 2 Boxplot 1 M1 M2 Boxplot 2
8 8 8
6 6 6 1 2 3 R R R
P 4 P 4 P 4 A A A
2 2 2
0 0 0 Boxplot 1 M1 M2 Boxplot 2 Boxplot 1 M1 M2 Boxplot 2 Boxplot 1 M1 M2 Boxplot 2
Figure .: Combined box-and-whisker and “Measurement - Measurement ” comparison scaerplots (joined with straight lines) showing arteriolar diameter fluctuation and responses (n=) across all three flickers. Outliers are depicted as dots. See Table . for numerical values. For acronyms, see page .
Retinal Vessel Analyser: Reproducibility
RT1 RT2 RT3 50 50 50 ) ) ) s s s ( ( ( t t t e e e s s s 40 40 40 n n n o o o r r r e e e 30 30 30 k k k c c c i i i l l l f f f e e e 20 20 20 c c c n n n i i i s s s e e e 10 10 10 m m m i i i T T T 0 0 0 Boxplot 1 M1 M2 Boxplot 2 Boxplot 1 M1 M2 Boxplot 2 Boxplot 1 M1 M2 Boxplot 2
CT1 CT2 CT3
) 50 ) 50 ) 50 s s s ( ( ( t t t e e e
s 40 s 40 s 40 n n n o o o r r r
e 30 e 30 e 30 k k k c c c i i i l l l f f f
e 20 e 20 e 20 c c c n n n i i i s s s
e 10 e 10 e 10 m m m i i i T 0 T 0 T 0 Boxplot 1 M1 M2 Boxplot 2 Boxplot 1 M1 M2 Boxplot 2 Boxplot 1 M1 M2 Boxplot 2
Figure .: Combined box-and-whisker and “Measurement - Measurement ” comparison scaerplots (joined with straight lines) showing arteriolar reaction and constriction times (n=) across all three flickers. Outliers are depicted as dots. See Table . for numerical values. For acronyms, see page .
... Comparison Between Inbuilt and Independent Analysis
To compare the inbuilt RVA soware analysis with the one independently calculated from raw data, the three flicker responses as per Table . were averaged. en, the two were compared using a parametric t-test. Results are shown in Table .. For both measurement sessions, all arterial and venular parameters calculated by means of the vendor-supplied RVA soware were underestimated, compared to the respective independently analysed ones. All except one (maximum dilation for measurement session ) reached statistical significance (p=.).
Parameter (%) Measurement p value Measurement p value A . (.) . () ︌︀︗ . . MD (arteries) . (.) . (.) A -. (.) -. (.) ︌︈︍ <. <. MC -. (.) -. (.) A . (.) . (.) ️︄︀︊ <. <. DA . (.) . (.) V . (.) . (.) ︌︀︗ <. <. MD (veins) . (.) . (.)
Table .: Comparison (n=) between inbuilt (see Table .) and independent flicker analysis (see Table .). Values are expressed as means (SD) % change to baseline diameter. Student’s paired t-tests were performed. Statistical significance is denoted in bold. For definitions, see Section ... on page .
Retinal Vessel Analyser: Reproducibility e e e g g g n n n 10 10 10 a a a h h h C C C r r r e e e t t t e e e m m m a a a i i i 5 5 5 D D D % % % 2 3 1 F F F D D D B B B 0 0 0 Boxplot 1 M1 M2 Boxplot 2 Boxplot 1 M1 M2 Boxplot 2 Boxplot 1 M1 M2 Boxplot 2 r r r e e e t t t
e 115 e 115 e 115 m m m a a a i i i D D D e e e
v 110 v 110 v 110 i i i t t t a a a l l l e e e R R R
% 105 % 105 % 105 1 2 3 D D D M M M 100 100 100 Boxplot 1 M1 M2 Boxplot 2 Boxplot 1 M1 M2 Boxplot 2 Boxplot 1 M1 M2 Boxplot 2
15 15 15
1 10 2 10 3 10 D D D a a a t t t l l l e e e
D 5 D 5 D 5
0 0 0
Boxplot 1 M1 M2 Boxplot 2 Boxplot 1 M1 M2 Boxplot 2 Boxplot 1 M1 M2 Boxplot 2
Figure .: Combined box-and-whisker and “Measurement - Measurement ” comparison scaerplots (joined with straight lines) showing venular diameter fluctuation and responses (n=) across all three flickers. Outliers are depicted as dots. See Table . for numerical values. For acronyms, see page .
RT1 RT2 RT3 50 50 50 ) ) ) s s s ( ( ( t t t e e e s s s 40 40 40 n n n o o o r r r e e e 30 30 30 k k k c c c i i i l l l f f f e e e 20 20 20 c c c n n n i i i s s s e e e 10 10 10 m m m i i i T T T 0 0 0 Boxplot 1 M1 M2 Boxplot 2 Boxplot 1 M1 M2 Boxplot 2 Boxplot 1 M1 M2 Boxplot 2
Figure .: Combined box-and-whisker and “Measurement - Measurement ” comparison scaerplots (joined with straight lines) showing venular reaction times (n=) across all three flickers. Outliers are depicted as dots. See Table . for numerical values. For acronyms, see page .
Retinal Vessel Analyser: Reproducibility
... Averaged Flier Responses
Similarly with results from the interobserver reproducibility analysis, since the majority of research groups are performing averaging across the three flicker cycles, we are also presenting reproducibility results of all previously calculated parameters, this time with averaged values, collapsing data across all flicker repetitions (Tables . to .). Averaged values followed a normal distribution, hence parametric tests were performed to check for differences.
... Static Retinal Vessel Parameters
CRAE, CRVE, AVR, tortuosity indices and branching angles values are shown in Table ..
... Multiple Regression Analysis
Forward stepwise multiple regression analysis was used to test if any of the static and dynamic parameters (averaged across flicker cycles) or any of the measured baseline vessel diameters and IOP or BP values significantly predicted MD (separately for arteries/veins and measurement sessions). For arteries, the dependent variable was arterial MD and the independent variables that were tested whether they predicted MD responses were: absolute arteriolar diameter, IOP, MABP, CRAE, arterial tortuosity, proximal and distal arteriolar bifurcation angles, BDF, bFR, MC, DA, ΔD, APR, RT and CT. For veins, the dependent variable was venular MD and the independent variables that were tested whether they predicted MD responses were: absolute venular diameter, CRVE, venular tortuosity, proximal and distal venular bifurcation angles, BDF, ΔD, and RT. For measurement , the results of the regression for arterial MD indicated that three predictors ,., F(,)=, p<.). ese were DA (β==ﺾ explained .% of the variance of MD (R p<.), MC (β=., p<.) and CT (β=-., p=.). For measurement , the results of the regression for arterial MD indicated four predictors that could explain .% of the variance ,., F(,)=, p<.). ese were ΔD(β=., p=.), bFR (β=.=ﺾ of MD (R p<.), MC (β=., p<.) and BDF (β=., p<.). For measurement , the results of the regression for venular MD indicated that two predictors ., F(,)=., p<.). ese were ΔD=ﺾ explained .% of the variance of MD (R (β=., p<.) and BDF (β=., p<.). For measurement , the results of the regression for venular MD indicated that the same two predictors explained % of the variance of ,., F(,)=., p<.). ese were ΔD(β=., p<.) and BDF (β=.=ﺾ MD (R p<.).
Retinal Vessel Analyser: Reproducibility
Measurement Measurement Parameter ICC p value Arterioles BDF (%) . (.) (.) . . MD (%) . (.) . (.) . . MC (%) . (.) . (.) . . DA (%) . (.) . (.) . . bFR (%) . (.) . (.) . . ΔD (%) . (.) (.) . . APR (.) (.) . . Venules BDF (%) . (.) . (.) . . MD (%) . (.) . (.) . . ΔD (%) . (.) . (.) . .
Table .: Independently analysed RVA dynamic flicker response parameters (n=) compared between measurement sessions, averaged across all three flicker cycles. Values are expressed as means (SD). Student’s paired t-tests were performed. For acronyms, see page .
Measurement Measurement Parameter ICC p value Arterioles RT (seconds) () () -. . CT (seconds) () () -. . Venules RT (seconds) () () . .
Table .: Independently analysed RVA dynamic response parameters (n=) compared between examiners, averaged across all three flicker cycles. Values are expressed as means (SD). Student’s paired t-tests were performed. For acronyms, see page .
Parameter Arteries Veins CRAE (μm) () N/A CRVE (μm) N/A () AVR . (.) Branching Angle (proximal) (degrees) . () . () Branching Angle (distal) (degrees) . () . () Tortuosity Index . (.) . (.)
Table .: Static retinal vessel parameters of the healthy, intraobserver cohort (n=). Values are expressed as means (SD). For acronyms, see page .
Retinal Vessel Analyser: Reproducibility
Lastly, the test of whether arteriolar MD, MABP and IOP predicted the extent of venular MD reached no significance for both measurement sessions.
. Discussion
.. Interobserver Reproducibility Study
e effect of the strict standardisation procedures during data collection by means of the RVA system is reflected through the high ICC values of absolute arterial and venular diameters selected by the two Examiners (Table .). ese confirm that both Examiners selected the same vessel segment. Hence, vessel reaction to flicker stimulation is highly unlikely to be confounded by any (potential) influence on initial absolute vessel diameter.
e inbuilt RVA dynamic response parameters, averaged across all three flicker cycles, exhibit excellent reproducibility across Examiners. As mentioned in Section ..., these values are calculated from an arbitrarily chosen time window, that encompasses seconds: seconds before flicker cessation and seconds aer. is means, that the maximal dilation and constriction are “expected” to take place within a fixed time frame, for each flicker cycle: - seconds aer flicker initiation. Recently, this assumption has proven problematic, as reaction times can vary outside this seconds time frame (Heitmar et al., ). In line with Heitmar and colleagues, when comparing the two different analyses, statistically significant differences were found for arterial MD and MC when compared with their counterparts (A︌︀︗ and A︌︈︍ respectively) (see Table .). erefore, independently analysing raw RVA data and reporting the maximal dilation and constriction values from a wider time window of seconds (aer flicker initiation) eliminates the underestimation or overestimation of flicker responses.
Interobserver reproducibility of the independently analysed parameters generally showed moderate ICC values, similarly for both arteries and veins (Table .). Only the DA parameter, relevant to arteries, showed excellent reproducibility results across all three flicker cycles. ese results might be explained by two factors: the relatively small sample size per Examiner (n=) and the inherently variable nature of retinal haemodynamics. Regarding comparisons across the three flicker cycles within Examiners, the non-parametric Friedman test revealed no statistically significant differences. is result gives support to the notion that no comparable differences exist within a single measurement across the three flicker repeats, therefore averaging values would make analysis less complicated without sacrificing information.
Reaction and constriction times for arteries and veins showed moderate reproducibility between Examiners, whereas values did not differ across the three flicker cycles within Examiners. Interestingly, for Examiner , veins needed significantly longer time than arteries to reach maximum dilation (Table .). is was the case for two out of three flicker cycles.
Retinal Vessel Analyser: Reproducibility
A similar “delay” of approximately - seconds of the venous reaction has been previously reported for healthy subjects (Heitmar et al., , b; Lanzl et al., ; Kotliar et al., b) in line to this finding. e fact that this finding was not evident for measurements taken from Examiner , might be explained from the relatively small sample size (n=, i.e. low statistical power). Performing an a priori power analysis for a two-tailed Wilcoxon signed-rank test (matched pairs) at an alpha level (α) of . by means of the G*Power soware (version ..) (Faul et al., ), revealed that a sample size (n) of would be required for a large effect size (.) with % statistical power, whereas for a medium effect size (.) the sample size would have to be increased to . Nevertheless, the range of values for reaction and constriction times reported here, are in very good agreement to the ones obtained from a sample of healthy South Asians and White Europeans (comparable to the sample characteristics of this study) (Patel et al., ). e majority of within flicker and across Examiners comparisons did not reveal statistically significant differences, with the exception of arterial RT of the second flicker cycle. is, in conjunction with the results mentioned above, might be proof that different Examiners with similar experience may yield comparable results using the RVA system.
.. Intraobserver Reproducibility Study
Having a considerably larger sample size of healthy volunteers (n= versus n=) for the intraobserver reproducibility study, it was possible to overcome the limitations of the interobserver reproducibility study. e baseline characteristics of this cohort (Table .) confirm the healthy status of the participants and that these adhered to the inclusion and exclusion criteria. Contrary to the comparison between Examiners, when all measurements took place on the same day, in this part of the study, measurements took place either on the same day or on a separate visit. Hence, it is important, that prior to both measurements, IOP and BP values showed no significant differences between the two measurement sessions, rendering all subsequent comparisons relevant.
Although, both arterial and venular absolute diameters between measurement sessions had highly reproducible values (shown by ICC values of more than .), the diameter of the selected venular segment was (on average) microns narrower during the repeat measurement (Table .). is difference might have arisen from measurements that took place on separate days, when the repetition feature of the soware could not be utilised and manual matching of the vessel segment was performed. Nevertheless, it has been reported that “baseline vessel diameter does not influence relative magnitude of the flicker response” (Gugleta et al., ). Assuming that this diameter difference was a defining factor in terms of flicker response, one would expect to find a significant difference in the venular maximum dilation values. In fact, when the two measurement sessions were compared, this was not the case, neither with the
Retinal Vessel Analyser: Reproducibility
V︌︀︗ parameter (Table .), nor with the MD parameter (Table .). Results, thus, are in line with previous findings (not influenced by baseline vessel diameter differences).
Similarly to the previous sub-study, the inbuilt RVA dynamic response parameters, averaged across all three flicker cycles, exhibit high reproducibility across measurement sessions. e maximum arterial constriction (A︌︈︍) is an exception, with a low ICC value of . (Table .). ough, as previously mentioned, the validity of the soware-generated parameters is more important in this case, rather than how much reproducible they are. As such, comparisons performed between the independently analysed reaction parameters and their counterparts re- vealed an even stronger difference (compared to the interobserver study) for both measurement sessions (Table .). Namely, the inbuilt analysis consistently underestimated all responses, confirming to a greater extent the same finding as in the smaller cohort of participants.
Regarding reproducibility of the independently analysed dynamic responses, of note is the excellent consistence of values relating to venular parameters (BDF, MD and ΔD) throughout the flicker cycles (Table .) between measurements. On the other hand, arterial parameters show moderate reproducibility in general compared to veins. is could be explained by considering some imaging aspects during data capturing: veins appear darker than arteries and thus exhibit higher contrast to their background. is, in turn, makes veins less susceptible to erroneous diameter estimations and vice versa. Comparisons within measurement sessions, across flicker repeats did not reveal statistically significant differences, with one exception: maximum venular dilation (for measurement ) showed an increasing trend from flicker to flicker (p=.).
How do the main outcome measures of this thesis (Table .) compare to the ones found in literature? Since there are few studies that have dealt exclusively with healthy volunteers, to answer this question, one must refer to publications including healthy populations as controls, compared across various pathological states. Arterial BDF and bFR have been reported once on a per flicker basis (Heitmar et al., ) in literature. Values fluctuate approximately % lower than the ones in the aforementioned study for the former, but are comparable for the laer. ere are no reports on venular BDF on a per flicker breakdown, to compare these findings to. On averaged venular BDF values, Patel et al. () report slightly higher values. Arterial and venular MD values are comparable with the ones reported by several studies (Nagel et al., a; Mandecka et al., ; Lasta et al., ), though these studies have averaged responses across flicker cycles. Approximately % lower maximal dilation and % larger maximal constriction was found among subjects - for all flickers - compared to the normals (of an older age group) included in Heitmar et al. (b). is translates in a comparable arterial DA, despite the age difference. e study (Heitmar et al., ) that introduced retinal arteriolar elasticity (defined
Retinal Vessel Analyser: Reproducibility as APR) showed similar values to the ones reported here. e index ΔD representing arterial and venular dilatory capacity shows comparable values between vessel types.
e findings of delayed venular RT compared to arterial RT in the interobserver study, were confirmed with greater power in the intraobserver study. When reaction times of arteries and veins where compared, values from all flicker cycles within the first measurement session reached statistical significance, whereas one cycle did so for the second measurement (the trend that venules took longer to reach maximum dilation, remained though) (Table .). Also it is important to note that both reaction and constriction times have low ICC values, indicating a large fluctuation between sessions and making it debatable whether they are meaningful to calculate. No significant differences were found within measurement sessions. Looking at reaction time values, it is evident that arteries may reach maximum dilation (on average) outside the - seconds window (for instance on the th second). is is essentially the source of underestimation when using the inbuilt RVA parameters. It is even more pronounced when calculating the maximum constriction, as constriction times are well outside this range (ranges of - seconds). Venular reaction times (on average) fall within the - seconds window, but not all individual values do so, thus rendering comparisons between the two analyses statistically significant (Table .). e values reported here are comparable with several other studies (Patel et al., ; Heitmar et al., , b) highlighting the importance and benefits of standardisation when comparisons across studies are to be performed.
Regression analysis did not reveal any relationship between static (baseline diameters, CRAE, CRVE, tortuosity index, bifurcation angles) and dynamic (MD) parameters, neither in arteries nor in veins. MABP did not predict arteriolar MD in any of the regression models, confirming other studies (Heitmar et al., ; Garhöfer et al., ). Interestingly, the two measurement sessions did not “agree” in which predictors could explain the variance of arterial MD. is is one more indicator of low agreement among arteriolar calculated parameters. e only common predictor was arteriolar MC, which shows a positive correlation with arteriolar MD. In other words, arteries responding to flicker with large dilation responses exhibit large constriction phases aer flicker cessation. Conversely, for veins the two predictors (BDF and ΔD) explained the variance of MD consistently throughout the two measurement sessions. By definition, ΔD is a measure of a vessel’s dilatory capacity, which explains its relation to MD as shown by the regression analysis results. Of note is that venular BDF, which represents the diameter fluctuation during baseline illumination, can predict MD variation during flicker stimulation. Spontaneous retinal venous pulsation is a well known observation occurring in the proximity of the ONH (Jacks and Miller, ). is might contribute to venous BDF despite the considerable distance of the measured vessel segment from the edge of the ONH. Results
Retinal Vessel Analyser: Reproducibility support the notion that large baseline fluctuations lead to large dilation responses during flicker stimulation.
. Conclusions
Despite the innovation the RVA system has brought into the field of non-invasive retinal function assessment, research groups do not follow standardised analysis procedures yet and sometimes fail to extensively describe their implemented methods and protocols. Reprodu- cibility of the flicker responses strongly depends on measuring conditions (Seifertl and Vilser, ). ese should be carefully replicated as carefully as possible, when comparisons are to be made between healthy participants and various disease populations, to ensure the validity of the measurements.
Herewith, results are presented of interobserver and intraobserver reproducibility of a series of parameters that exist to describe the arterial and venular compliance before, during and aer flicker provocation. Despite the strict standardisations applied throughout, ICC values are low to moderate for arteries, whereas for veins are moderate to high. Reproducibility is helpful to examine when there is no satisfactory standard against which to compare the validity of a measurement. Studies from other research groups are warranted to compare these results to.
In CVD the delicate balance between vasodilators and vasoconstrictors is disturbed leading to what is commonly referred to as endothelial dysfunction (Nadar et al., ). Hence, accurate assessment of vascular function has been investigated as a potential prognostic marker and as a possible therapeutic target. Currently, the golden standard to assess endothelial (dys)function in a non-invasive manner and on a macrovascular level is Flow-Mediated Dilation (FMD). Evaluation of FMD in the brachial artery is performed by means of high-resolution ultrasound recording the physiological response of increased blood flow, following distal forearm induced ischemia. A recent, multi-center reproducibility study has demonstrated that adherence to a rigorous protocol and adequate operator skills improve the reliability of the technique (Ghiadoni et al., ). Short-term CVs were ranging from .% to .%, whereas long- term CVs were ranging from .% to .%. Flicker-induced vessel responses were weakly correlated to brachial FMD indicating the absence of a direct analogy between the two vascular beds and possibly between the two mechanisms involved (Pemp et al., ).
Current studies employing the RVA system to assess endothelial function are pursuing to explore potential flicker response differences between health and disease. Since these comparisons are naturally cross-sectional, the applicability of the RVA is currently limited to screening or stratification purposes. Lack of longitudinal studies is preventing researchers to
Retinal Vessel Analyser: Reproducibility draw conclusions on the progression of the associations reported. Such studies are needed to explore whether the RVA can be used as a diagnostic tool in the future.
Chapter
Location and Length Influence on Vessel Reactivity
. Baground
Documents provided by the manufacturer of the RVA system (Imedos AG) describe certain technical limitations as well as certain standardisation procedures which should be followed by end-users. More specifically, on selecting the measurement location and vessel segment’s length, the following are advised:
• Measurement location should be at least . DD away from the ONH.
• Vessel segment’s length should have a maximum length of DD, but generally recom- mended to be kept as long as possible.
• Due to resolution limitations, measurement of vessels with luminal diameter smaller .”μm “may be difficultﺼﻅ than
e measurement setup procedure encompasses the following steps. Initially, the camera’s objective is adjusted centrally to the dilated pupil to an appropriate distance in order to obtain a uniformly illuminated fundus image on the computer’s screen. en, focus is adjusted to get a sharp image. e fixation needle needs to be placed accordingly so as the subject is able to clearly observe it and at the same time the vessels of interest are central to the image view. e end-user finally places the measurement window on the desired area. An instance of the measuring window is shown in Figure .. e pair of red lines superimposed on an arteriole and a venule indicate the measurement location and length.
How do the aforementioned manufacturer guidelines translate into numerical values? Typic- ally, in a healthy eye, arterioles’ diameters are ranging from μm to μm, whereas venules’ diameters from μm to μm, within a range of - DD from the edge of the ONH. Hence, there is a substantial range of vessel diameters (i.e. different locations) that can be measured.
Location and Length Influence on Vessel Reactivity
Regarding vessel’s length, theoretically, the maximum length, that the red lines in Figure . ︔︒︋️ﺼﻁﻀcan be extended to, is pixels (personal communication). Using the Carl Zeiss FF fundus camera at the ° angle image field, these correspond to MU or else to μm for Gullstrand’s normal eye (. MU per pixel). Practically, this maximum value cannot possibly be achieved, since the soware’s algorithm truncates parts of the selection edges, even under ideal conditions. For example, the maximum length that could possibly be selected using a stationary straight target (for testing purposes) yielded a value of MU.
.. Motivation and Resear Rationale
In addition to the lack of standardisation on dilatory parameter calculations (Chapter ), currently there is no standardisation regarding the location and the length of the vessel segments selected for retinal endothelial functional assessment by means of the RVA. Although the majority of research groups appear to sample the retina within - DDs away from the ONH, many others arbitrarily select within a wide range of - DDs (Nagel et al., a,b; Frederiksen et al., ; Bek et al., ; Mehlsen et al., ). Furthermore, even though the majority of researchers report the location they measured at, they do so without applying any standardisation, but base their selection on visual estimations only. Some do not define their selection at all (Blum et al., ; Rueddel et al., ).
Similarly for the length of the chosen vessel segments, many studies do not disclose any information (Rickenbacher et al., ; Reimann et al., ; Pemp et al., ; Lo et al., ; Lasta et al., ), whereas some that do so, report values as long as μm (Mandecka et al., ; Nguyen et al., ; Dawczynski et al., ; Mandecka et al., ) that by definition cannot be true (longer than the theoretical upper limit of μm). Moreover, no study reports numerical values of absolute vessel segment lengths, despite the fact that this information can be easily extracted from the supplied soware. Instead, the values that are reported are qualitative approximations. Also, it is questionable whether values reported are valid, because finding a long straight vessel segment (of an extent of μm) on all measured retinas in a study is improbable.
Understandingly, both vessel location and length selection are governed by individual angioar- chitecture. But within the ° angle image field, retinal blood vessels vary structurally and functionally, as a function of size and location. e use of relative (to the baseline) diameter values to flicker reaction helps to overcome vessel size differences, but vessel dilation and/or constriction might be influenced by locality and/or the extent of segments measured. Herewith, it is investigated whether these two variables - location and length - affect the measuring outcomes.
Location and Length Influence on Vessel Reactivity
Figure .: RVA’s measuring window. e four blue circles at each corner aid the repositioning of the fixation needle (shown at right) for repeated measurements. Here, the superotemporal fundus area of a le eye is shown.
. Subjects and Methods
Measurements were performed by a single examiner for healthy volunteers ( males, females) in one unselected eye. e same inclusion and exclusion criteria as previously listed (Section ..), applied.
.. Data Collection
e same protocol was followed as in Section ... using three flicker cycles as in Figure .. For every given RVA measurement session, vessel diameters of one pair (arteriole-venule) of operator-selected vessel segments (within an area of - DD) were recorded across space and time, in real time. ese sessions were recorded on S-VHS tapes. is allowed us to replay sessions offline and select two additional measurement locations (- DD and - DD) as per Figure .. Measurement rings as the ones shown in Figure . were superimposed on each measurement window according to every individuals’ DD to improve standardisation and achieve perfect measurement location accuracy. e use of relative distances to the ONH to describe retinal locations is a standard procedure, as for AVR measurements, for instance. e choice of DD wide rings, on one hand serves for easy comparisons among different groups (because this is what most groups report) and on the other hand satisfies the manufacturer’s recommendation for taking vessel segments of a maximum of DD length. Despite the fact that the length of each vessel segment was arbitrarily selected (within the limits of each ring)
Location and Length Influence on Vessel Reactivity prior to the commencement of the measurement session, this was later standardised across subjects, as detailed in the next Section.
.. Data Processing
Data were analysed independently of the vendor-supplied soware. Custom-built scripts were used to process raw data output from the RVA soware in a versatile way. ese scripts were wrien by Dr. Robert J. Summers. Excel data matrices containing both the artery and vein flicker reaction across space (one diameter recording every MU, extending to a variable amount of columns, depending on segment length) and time ( diameter recordings per second, for seconds equalling to a total maximum of rows of diameter data) were split into two Comma-separated Values (CSV) files using a bash script, by means of the xlscsv and csplit utilities. Further awk scripts:
• removed any potential outliers (eliminating irregularly high or low values due to e.g. (SDs from the mean, similarly to others (Jensen et al., ﺾ± blinks), data outside • filled in any potential missing data blocks, using linear interpolation, in line with Kotliar et al. (a)
• binned values over second intervals (each containing data points), in line with Nagel et al. (); Gugleta et al. ()
• normalised vessel diameter values to the mean of the seconds of baseline diameter values, in line with Polak et al. (); Kotliar et al. ()
... Analysis per Location
Further to the real-time recordings of one arteriole and one venule within the area of - DD (designated as Segment ) distal to the ONH, two additional pairs of vessel segments were measured offline from the video tape recordings. ese were within a range of - DD (designated as Segment ) and - DD (designated as Segment ) distal to the ONH. Hence, a total of arteriolar vessel segments and venular segments went into the analysis. To eliminate any potential bias of vessel length selection, an equally long vessel segment ( MU, corresponding to columns of data), was processed for all segments across subjects with the aid of the aforementioned scripts. Absolute arteriolar and venular diameters were recorded in MU and compared across measurement locations.
Location and Length Influence on Vessel Reactivity
Figure .: Illustration of measured locations in relation to the distance from the ONH. e segment taken out from the area of - DD distal to the ONH is designated as Segment , from the area of - DD as Segment and from the area of - DD as Segment .
... Analysis per Segment’s Length
For the purpose of comparing vessel diameter’s flicker responses of different lengths all vessel segment pairs measured from the - DD area were re-analysed. During that re- analysis, instead of truncating the vessel lengths to MU, their full length was utilised as initially selected during the real-time measurement session. Depending on individual angioarchitecture, these lengths ranged from MU to MU for arteries and from MU to MU for veins. Detailed breakdown on a per individual basis is shown in the Results section (Table .). e diameter recordings during flicker stimulation of these longer segments went into a comparison with the previously standardised lengths of MU.
.. Data and Statistical Analysis
SPSS (Version . Chicago, SPSS Inc.) was used for statistical analysis and for ploing purposes. Normality tests were performed on all continuous data by means of the Shapiro-Wilk test, to determine distribution. For the case of absolute vessel diameters, values were normally distributed, thus one-way Analysis of Variance (ANOVA) was used with the measurement location as the categorical independent variable and vessel diameter as the continuous variable. For all other outcome parameters calculated, which were non-normally distributed, Kruskal- Wallis H tests were performed, with the measurement location as the grouping variable. For
Location and Length Influence on Vessel Reactivity all calculations, a P value of < . was considered significant. Associations of the outcome parameters (where applicable) with vessel diameter and MABP were examined by means of linear regression analysis. Finally, a graphical representation for comparing long and short segments was used by ploing the difference of the vessel diameter flicker response of each of the longest vessel segments minus their MU long counterparts.
. Results
.. Baseline Characteristics
Data from healthy volunteers ( males, females) were analysed and included in this study. Values of baseline parameters (vessel diameters across all measured segments, age, SBP, DBP, MABP, HR and IOP) are shown in Table .. A one-way ANOVA was used to test for vessel diameters differences among the three measurement locations. Values did not differ significantly across the three measurement locations, neither for arterioles (F (, ) = ., p = .), nor for venules (F (, ) = ., p = .).
.. Comparison Across Vessel Segments
e following parameters (as defined in Section ...) were calculated and compared (across three measurement locations) for all three flicker cycles by means of raw RVA output data processing: BDF, bFR, MD, MC, DA, RT, CT, ΔD and APR. Kruskal-Wallis H tests showed no statistically significant differences across any of the RVA dynamic flicker responses between the three measurement locations: neither in a flicker per flicker analysis (Table .), nor when all flicker cycles were averaged together (Table .).
Segment Segment Segment Parameters p-value (- DD) (- DD) (- DD) Arteriolar Diameters (MU) () () () . Venular Diameters (MU) () () () . Age (years) () SBP (mmHg) () DBP (mmHg) () MABP (mmHg) () HR (pulses/min) () IOP (mmHg) ()
Table .: Absolute arteriolar and venular diameters (n=) across three measurement locations and baseline characteristics of the cohort. Values are expressed as means (SD). For acronyms, see page .
Location and Length Influence on Vessel Reactivity
Arterioles Parameter Segment Segment Segment p value (- DD) (- DD) (- DD) BDF (%) . (.) . (.) . (.) . BDF (%) . (.) . (.) . (.) . BDF (%) . (.) . (.) . (.) . MD (%) . (.) . (.) . (.) . MD (%) . (.) . (.) . (.) . MD (%) . (.) (.) . (.) . MC (%) . (.) (.) (.) . MC (%) . (.) . (.) . (.) . MC (%) (.) . (.) (.) . DA (%) . (.) . (.) . (.) . DA (%) . (.) . (.) . (.) . DA (%) . (.) . (.) . (.) . bFR (%) . (.) . (.) (.) . bFR (%) . (.) . (.) . (.) . bFR (%) . (.) . (.) . (.) . ΔD (%) . (.) . (.) . (.) . ΔD (%) . (.) . (.) . (.) . ΔD (%) . (.) . (.) . (.) . APR . (.) . (.) . (.) . APR . (.) . (.) . (.) . APR . (.) . (.) . (.) . Venules BDF (%) . (.) . () . (.) . BDF (%) . (.) . (.) . (.) . BDF (%) . (.) . (.) . (.) . MD (%) . (.) . (.) . (.) . MD (%) . (.) () . (.) . MD (%) . (.) . () . (.) . ΔD (%) . (.) . (.) . (.) . ΔD (%) . (.) . (.) . (.) . ΔD (%) . () . (.) . (.) .
Table .: RVA dynamic flicker response parameters across three fundus locations (n=) in a flicker by flicker analysis. Values are expressed as medians (IQR). Kruskal-Wallis H tests were performed for comparisons across measurement locations. For acronyms, see page .
Location and Length Influence on Vessel Reactivity
Arteriolar reaction times showed a trend of slower reaction (time needed to reach maximum dilation) for larger vessels (closer to the ONH) compared to smaller ones (further away from the ONH). is was the case when individual flicker analysis was performed (Table .) and when flickers where averaged (Table .). A Kruskal-Wallis test revealed that this trend reached statistical significance in the case of averaged flickers (Table .): there was a statistically significant difference between the different measurement locations (H(,n=) = ., p = .), with a mean rank of . seconds for Segment , . seconds for Segment and . seconds for Segment . A post-hoc test using Dunn’s multiple comparisons showed a significant difference in arteriolar reaction times between Segment and Segment (p = .). Segment compared to Segment followed a similar trend, however failed to reach statistical significance (p = .).
Since the previous non-parametric Kruskal-Wallis H tests showed no significant differences across the three measurement locations for all outcome parameters (as shown in Table .), all measurement sites (n=) were pooled together for linear regression analysis. Exploring potential associations of the outcome parameters with absolute vessel diameters did not show statistically significant correlations. For example, neither absolute arteriolar diameter (r=-., p=.), nor absolute venular diameter (r=-., p=.) correlated with maximum arteriolar and venular dilation, respectively. On the contrary, there was a significant positive correlation between MABP of subjects and arteriolar diameter response induced by flickering light (r=., p=.; Figure .).
.. Flier Responses Variability as a Function of Location
To assess whether the measurement location influences the variability of the outcome meas- ures, coefficients of variation were calculated across all three flicker cycles and then averaged across participants (n=). Results for both arterioles and venules are summarised in Table ..
.. Comparison Between Segment Lengths
Output from continuous diameter recordings as per the standard flicker protocol were extracted from the same vessel selection, twice: once for the longest possible selection and a second time from the truncation of this long segment to a MU long segment. e point by point subtraction of each diameter value derived from the long and short segments are ploed in Figure . for arteries and Figure . for veins, for all subjects (n=). e analysis was constrained only within a - DD range from the edge of the ONH (i.e. only “Segments ” were analysed).
Location and Length Influence on Vessel Reactivity
Arterioles Parameter Segment Segment Segment p value (- DD) (- DD) (- DD) RT (seconds) () () () . RT (seconds) () () () . RT (seconds) () () () . CT (seconds) () () () . CT (seconds) () () () . CT (seconds) () () () . Venules RT (seconds) () () () . RT (seconds) () () () . RT (seconds) () () () .
Table .: RVA dynamic flicker reaction and constriction times across three fundus locations (n=) in a flicker by flicker analysis. Values are expressed as medians (IQR). Kruskal-Wallis H tests were performed for comparisons across measurement locations. For acronyms, see page .
Arterioles Parameter Segment Segment Segment p value (- DD) (- DD) (- DD) BDF (%) . (.) . (.) . (.) . MD (%) . (.) (.) . (.) . MC (%) . (.) () . (.) . DA (%) . (.) . (.) . (.) . bFR (%) . (.) . (.) . (.) . ΔD (%) . () . (.) . () . APR . (.) . (.) . (.) . Venules BDF (%) . (.) . (.) . (.) . MD (%) . (.) . (.) . (.) . ΔD (%) . (.) (.) . (.) .
Table .: RVA dynamic flicker response parameters across three fundus locations (n=), averaged across three flicker cycles. Values are expressed as medians (IQR). Kruskal-Wallis H tests were performed for comparisons across measurement locations. For acronyms, see page .
Location and Length Influence on Vessel Reactivity
Arterioles Parameter Segment Segment Segment p value (- DD) (- DD) (- DD) RT (seconds) ()† ()† () . CT (seconds) () () () . Venules RT (seconds) () () () .
Table .: RVA dynamic flicker reaction and constriction times across three fundus locations (n=), averaged across three flicker cycles. Values are expressed as medians (IQR). Kruskal-Wallis H tests were performed for comparisons across measurement locations. Statistical significance is denoted in bold. † signifies post-hoc test’s statistically significant differences between Segment and Segment (p = .). For acronyms, see page .
Figure .: Correlation between the maximum arteriolar dilation response and MABP for all segments () of all subjects (n=). e solid line corresponds to the regression line and the dashed line corresponds to the % confidence interval. Segment corresponds to an area of - DD, Segment to an area of - DD and Segment to an area of - DD distal to the ONH.
Location and Length Influence on Vessel Reactivity
Arterioles Parameter Segment Segment Segment (- DD) (- DD) (- DD) BDF .% .% .% MD .% .% .% MC .% .% .% DA .% .% .% bFR .% .% .% ΔD .% .% .% APR .% .% .% RT .% .% .% CT .% .% .% Venules BDF .% .% .% MD .% .% .% ΔD .% .% .% RT .% .% .%
Table .: Mean coefficients of variation across flicker cycles as a function of location for each outcome parameter. Smallest variability per parameter is highlighted in bold. For acronyms, see page .
Arteries Veins Subjects Short Long Difference Short Long Difference MU MU MU MU MU MU
Table .: Vessel segments lengths that went into the comparison. e extent of the long segments was only restricted by individual angioarchitecture (selection was always kept as long as possible). e extent of the short segments was truncated to an arbitrary length of MU in order to compare two distinctly different lengths.
Location and Length Influence on Vessel Reactivity
To quantify the graphical representation of the differences as shown in Figures . to ., the area under the curve (positive deviation from zero), the area above the curve (negative deviation from zero) and their sums (total deviation from zero) were calculated (Table .) for arteries and veins individually per subject. Since the subtraction was (arbitrarily) calculated as the long segment minus the short one, larger positive deviations from zero compared to their negative counterparts means that longer segments showed higher amount of dilation (on average, across the seconds) compared to the shorter segments and vice versa. For arteries, the best agreement (i.e. total deviation values closer to zero) between the two lengths selections is found for Subjects , and . For veins, the best agreement between the two lengths selections is found for Subjects , and . is leaves the majority of participants ( out of ) with considerable deviations from the theoretical value of zero, indicating an influence of vessel segment’s length on diameter recordings.
No correlation was found (neither for arteries, nor for veins) between the absolute length difference between the long and the short vessel segments (Table .) and the induced total deviation from zero (Table .).
. Discussion
.. Measurement Location Considerations
Retinal vessels have varying structural and functional properties as they extend along space. Smooth muscle cells provide structural support to the vasculature and mediate the myogenic mechanism for vascular autoregulation of blood flow. As shown by electron microscopy, the arterial wall consists of five to seven layers of smooth muscle cells (tunica media) near the optic disk (Pournaras et al., ). Towards the equator, these decrease to two or three layers. An additional variable parameter when examining different locations along the retina is vessel branching. Depending on individual angioarchitecture an arteriole belonging to the area of - DD away from the edge of the ONH may belong to a third order bifurcation, since it is quite common for feeding arterioles to branch off towards the macula in the area prior to that (- DD). Consequently, there is a pressure drop between arterioles of differing branching order, which may affect the potential of each vessel segment to dilate, accordingly (Gafiychuk and Lubashevsky, ). is is corroborated from the fact that axial velocity of red blood cells in the major retinal arteries and veins of normal human and primate monkey eyes has been found to increase linearly with vessel diameter by means of bidirectional laser Doppler velocimetry (Pournaras et al., ).
Regarding functional heterogeneity along the retinal microvasculature structural differences on a cellular level in endothelial cells have been identified in the pig retina in vitro by means of
Location and Length Influence on Vessel Reactivity 350 350 350 350 350 300 300 300 300 300 250 250 250 250 250 200 200 200 200 200 Time (sec) Time (sec) Time (sec) Time (sec) Time (sec) Time Legend 150 150 150 150 150 100 100 100 100 100 Line indicating the subtraction's result (0) if no segment length in fl uence existed Point byPoint point subtraction of the responses fl icker (long minus short vessel segment) 50 50 50 50 50 Subject 2 Subject 4 Subject 6 Subject 8 Subject 10 Subject
5 0 5 0 5 0 5 0 5 0 -5 -5 -5 -5 -5
Long - short segment (MU) segment short - Long (MU) segment short - Long (MU) segment short - Long (MU) segment short - Long (MU) segment short - Long 350 350 350 350 350 350 300 300 300 300 300 300 250 250 250 250 250 250 200 200 200 200 200 200 Time (sec) Time Time (sec) Time (sec) Time (sec) Time (sec) Time (sec) Time 150 150 150 150 150 150 100 100 100 100 100 100 50 50 50 50 50 50 Subject 9 Subject Subject 1 Subject 3 Subject 5 Subject 7 Subject 11 Subject
5 0 -5
10 5 0 5 0 5 0 5 0 5 0 -5 -5 -5 -5 -5
Long - short segment (MU) segment short - Long (MU) segment short - Long (MU) segment short - Long (MU) segment short - Long (MU) segment short - Long (MU) segment short - Long
Figure .: Plots of the flicker response difference between the longest possible segment measured and a truncated shorter segment of MU for all subjects for arteries. Vessel segments belonged to an area of - DD distal to the ONH.
Location and Length Influence on Vessel Reactivity 350 350 350 350 350 300 300 300 300 300 250 250 250 250 250 200 200 200 200 200 Time (sec) Time (sec) Time Time (sec) Time (sec) Time (sec) Time Legend 150 150 150 150 150 100 100 100 100 100 Line indicating the subtraction's result (0) if no segment length in fl uence existed Point byPoint point subtraction of the responses fl icker (long minus short vessel segment) 50 50 50 50 50 Subject 8 Subject 10 Subject Subject 2 Subject 4 Subject 6 Subject
5 0 5 0 -5 -5
15 10 10 5 0 5 0 5 0 -10 -5 -5 -5
Long - short segment (MU) segment short - Long (MU) segment short - Long (MU) segment short - Long (MU) segment short - Long (MU) segment short - Long 350 350 350 350 350 350 300 300 300 300 300 300 250 250 250 250 250 250 200 200 200 200 200 200 Time (sec) Time (sec) Time (sec) Time Time (sec) Time (sec) Time (sec) Time 150 150 150 150 150 150 100 100 100 100 100 100 50 50 50 50 50 50 Subject 1 Subject 7 Subject 11 Subject Subject 3 Subject 5 Subject 9 Subject
5 0 5 0 5 0 -5 -5 -5
10 10 5 0 5 0 5 0 -10 -10 -5 -5 -5
Long - short segment (MU) segment short - Long (MU) segment short - Long (MU) segment short - Long (MU) segment short - Long (MU) segment short - Long (MU) segment short - Long
Figure .: Plots of the flicker response difference between the longest possible segment measured and a truncated shorter segment of MU for all subjects for veins. Vessel segments belonged to an area of - DD distal to the ONH.
Location and Length Influence on Vessel Reactivity
Arteries Veins Subjects Deviation from zero Positive Negative Total Positive Negative Total . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Table .: Values of positive, negative and total deviation from the theoretical value of zero, if the long and short segments had no influence on output of diameter recordings for arteries (Figure .) and veins (Figure .). Essentially, positive values represent the area under the curve, negative values the area above the curve and total deviation their sum. confocal microscopy (Yu et al., ). e distribution of F-actin, endothelial cell size and shape, nucleus size, shape and position within the cell were determined as a function of location along the vascular tree.
Exploration of retinal vascular reactivity over a range of locations is not a recent concept. Despite the lack of cuing edge technology at the time, Lanigan et al. () demonstrated the variability of retinal vessel responses at different sites using isometric muscle contractions as a stimulus. Responses differed by at least % in various arteriolar sites and at least % in various venular sites.
Prior to the incorporation of the embedded flicker module in the RVA at a flicker frequency of . Hz, flicker stimulation was achieved using an external lamp, which shone light onto a rotating sector disc (Polak et al., ). In their study, their main motivation was to investigate the optimisation of the diameter response in terms of different flicker frequencies (- Hz). At the same time though, they measured offline multiple locations using the video recordings: a major vessel trunk proximally (- DD), the same major vessel trunk distally (- DD) and a distal branch. Using two different flicker frequencies ( and Hz), they found comparable diameter flicker responses for arteries across all locations. In retinal veins, the response in branches (i.e. the smaller venules) had a tendency to be higher, but this effect was not significant for their healthy cohort of nine subjects. Results in this work, although, not directly comparable due the protocol differences, are in agreement, since no differences were found in arteriolar or
Location and Length Influence on Vessel Reactivity venular responses across the three different locations. ey also observed negative correlation between baseline diameters and maximum dilation values in veins, but not in arteries. Here, results showed a similar negative trend, but this was very weak and did not reach statistical significance.
In a sample of healthy, non-vasospastic females, Gugleta and colleagues found no significant differences in maximum dilation amplitude between proximal (- DD) and distal (- DD) retinal vessels (arteries and veins), with the three flicker cycles being analysed separately (Gugleta et al., ). Similarly to the results reported in this thesis, they reported no influence of the measurement site on vessel responses.
Another study investigated the influence of measurement location in healthy males in arterial diameter response by means of the RVA (Jeppesen et al., ). eir participants were subjected to isometric exercise, which induced an increase to systemic BP, while arteriolar diameters were continuously recorded. While this was a completely different experimental protocol (pressure autoregulation) compared to that of flicker stimulation (metabolic autore- gulation), which induces arterial contraction instead of dilation, it is of interest to note their findings: distal retinal arterioles (of smaller calibre) contracted significantly more than their proximal counterparts.
A recent study investigated differences in the response of arterioles supplying two different areas; the macular and the peripheral retina (Jensen et al., ). ey applied three different provocation protocols: isometric exercise, flickering light and a combination of the two. With flickering light alone, no differences were observed between the response in macular and peripheral arterioles within their healthy subjects.
e findings in this thesis extend current knowledge and certain methodological standardisa- tions are proposed. To implement the location-dependent comparisons, measurement rings analogous to the concept of AVR measurement rings (Figure .) were introduced. In this way, the three measurement sites are fully standardised across subjects, rather than using visual judgement to describe the approximate measurement location as the case is with current studies. For instance, Gugleta et al. () clearly show an example of vessel location selection to belong to an area of - DD, but it is not clear whether this location was kept for all of their subjects, since they mention that “vascular geometry governed the exact location of the measurement”. Moreover, spliing the areas in DD wide rings (- DD, - DD and - DD) fulfils the manufacturer’s recommendation to obtain vessel segments of a maximum length of DD.
e main outcome parameter, maximum arteriolar and venular dilation in response to flicker- ing light, did not differ significantly across the three measuring sites. is was the case both for
Location and Length Influence on Vessel Reactivity individual flicker analysis and for averaged flicker responses. e same applied for BDF, MC, DA, bFR, ΔD, APR and CT. Statistically, this is a direct implication that one could select any measurement location within a DD radius from the edge of the ONH without this having any impact on outcome parameters. e only outcome parameter that is an exception, according to results herein, is arteriolar reaction time between Segment and . When the three reaction times were averaged, proximal arterioles were slower to reach maximum dilation compared to their immediate neighbouring arterioles. ere are no other studies to have measured within a - DD area and to report reaction time values to compare this finding to.
Present data lend further support to previous observations (Polak et al., ; Nagel et al., ) that baseline absolute vessel diameters do not correlate with relative amplitude of flicker response, as confirmed by linear regression analysis, pooling all vessel segments together. Also, a positive correlation between MABP and maximum arteriolar dilation was found. A mmHg difference in MABP showed approximately a % difference in maximum arteriolar dilation induced by flicker provocation (within a range of normal BP values). Of course, correlation does not imply causation and since there are no relevant data in literature to compare this finding to, no further assumptions can be made at this point. A comparison between normotensives with MABP of mmHg and hypertensives with MABP of mmHg, showed an inverse relationship of diminishing maximum arteriolar dilation (.% versus .%, respectively), but used a longer protocol of flicker repeats (Nagel et al., ).
Despite the aforementioned structural and functional heterogeneities along the retinal mi- crovasculature, measurements of vascular reactivity in three distinct locations in arterioles and venules did not differ. ere could be various reasons for that. First, the vessels’ characteristics may not vary enough within - DD for the RVA’s resolving capabilities to be able to capture these differing vessel properties. Second, retinal diameters are inherently fluctuating even under constant illumination during the cardiac cycle. Reliable detection of small diameter changes (.%) was only possible by taking fundus photographs synchronised to an electrocardiograph, while other methods either failed to detect changes or were unreliable (Dumskyj et al., ). As the pulse wave travels along the microvasculature, flicker initiation may coincide with the crest, the trough or anywhere in between of the wave at a given vessel segment. erefore, responses may be blunted or augmented accordingly, masking any potential differences across different measuring locations. Lastly, the sample size might not suffice to detect significant differences. By all means, further studies on this topic are warranted to explore the influence of measurement location on retinal vessel reactivity by means of flicker stimulation.
Location and Length Influence on Vessel Reactivity
.. Measurement Length Considerations
Selecting the extent of the measuring vessel segment using the RVA system is primarily governed by individual angioarchitecture. Tortuous retinal vessels, bifurcations and closely situated arteries and veins are segments that cannot be considered for inclusion. Given that the soware’s algorithm automatically truncates parts of the segment selection in case of quality issues (for instance, low contrast between vessels and surrounding tissue), end-users should always aempt to select segments that are as long as possible.
For the first time, results on the effect of different vessel segment lengths on flicker provocation diameter recordings of both arterioles and venules are reported. So far, a large body from the RVA-related publications has given lile or no aention to systematic reporting of the extent of their vessel selection. From the analysis in this thesis, it is evident that for the majority of cases ( out of for arteries and out of for veins) the measurement length is a factor that induces variability to the final outcome. For both arteries and veins, there was substantial variability in the diameter responses when two distinctly different in length vessel segments were compared. is implies that including additional (or less) “information” from adjacent locations, yields a different diameter recording. One could argue that fluctuations are a consequence of noise or insufficient measurement quality (for instance, due to low contrast). However, this is highly unlikely, since all recordings were carefully selected prior to inclusion in this analysis. As a maer of fact, this is the reason for the relatively small sample size: only measurements with even illumination across the full ° field and high quality recordings could be fully analysed up to the - DD measurement ring. is was achieved by extracting the brightness course graphs by means of the vendor-supplied soware and validating that fluctuations were kept to a minimum along the measurement duration. It is also known, that optical distortions may add up to errors of measuring sensitivity up to % if the measuring location is located near the margins of the image area (Seifertl and Vilser, ). Hence, for added confidence, only vessel segments belonging to the central - DD measurement ring were included into the analysis. Another interesting observation is that in the case of veins (specifically for Subjects and ) the long segment selections exhibit consistently higher amounts of dilation (since the subtraction of long minus short selections are above zero) due to flickering light (time ranges of -, - and - seconds) by a factor of - MU.
Nevertheless, the implications of this novel finding do not necessarily have a negative impact to the reliability of the RVA system, if certain standardisations are enforced. Firstly, all relevant studies should be reporting the actual measurement location and the exact vessel length of their analysis. Unfortunately, a considerable amount of publications so far have not disclosed any such information. Secondly, comparisons within subjects (for example, from multiple
Location and Length Influence on Vessel Reactivity visit measurements) should always be made with equally long segment selections. Of course, this is practically not feasible to achieve in real time, but truncating offline vessel segments accordingly, prior to data analysis, can be easily accomplished.
. Conclusions
e vulnerability of the retina to vascular related diseases and the substantial reliance on local regulation of the retinal vasculature renders an improved understanding of such local regulatory mechanisms of significant clinical importance. Multi-segment analysis may indeed show comparable responses in healthy volunteers, but this may not be the case in various retinal manifestations of systemic disease, including diabetic retinopathy, glaucoma and hypertensive retinopathy.
Standardisation of measurement conditions is a necessity when utilising the RVA system. e magnitude of the induced responses by means of retinal flicker provocation is small and many factors can potentially suppress or augment those responses. Both measurement location and the extend of the vessel segment sampled should be taken into account to be able to control for these factors and should always be reported in future publications. Agreement between research groups on standardisation protocols needs to be reached, before the RVA can be considered clinically useful in detecting or predicting vascular dysfunction.
Chapter
Essential Hypertension: Case Reports
. Introduction and Motivation
Imaging the retinal microvasculature offers a surrogate view of systemic vascular health, allowing non-invasive and longitudinal assessment of vascular pathology. In order to discuss the strengths and weaknesses of utilising the RVA system to assess metabolic autoregulation in the retina in treated essential hypertensives three essential hypertensives were invited that were previously ( years ago) subjected to the protocol detailed below, as a follow-up, longitudinal, small case report study.
. Ethical Approval
e study adhered to the tenets of the declaration of Helsinki and the protocol was peer reviewed from Aston University as well as undergone through a separate second peer review by the Aston Optometry and Audiology Research Ethics Commiee, which subsequently approved it. Furthermore, this study has undergone R&D and NHS ethics review prior to its commencement (Research Ethics Commiee Reference: /EM/).
. Methods and Subjects
One previously diagnosed and two newly diagnosed essential hypertensives had been initially examined as part of a research study in and returned for a follow-up examination in . Both the initial and the follow-up assessments were split into two research appointments, held on two consecutive days as follows.
.. Day - Ambulatory BP and ECG Monitoring
Participants were invited to aend their first research appointment aer fasting from midnight of the preceding day. A hour BP and ECG monitor (Cardiotens, Meditech, PMS Instruments,
Essential Hypertension: Case Reports
UK) was fied to assess both systemic circulation and autonomic function. BP measurements were obtained every minutes during the day period and every minutes during the night period. Given - for instance - a typical hour sleep period the total number of BP measurements obtained amounted to . Patients recorded a standardised patient diary on the monitoring day with information on their daily routine, physical activities undertaken and time and type of antihypertensive medication taken. en, subjects were dismissed.
... Outcome Measures
At completion of the hour period the BP and ECG monitor was removed. Data were downloaded onto a personal computer and were analysed using the vendor-supplied soware, namely CardioVisions (Version ..). Outcome variables were SBP, DBP, HR, LF, HF, HRV triangular index, each for day, night and hour periods.
.. Day - Eye examinations
e following day, patients returned for their second research appointment. At least hours prior to their morning visits, participants were asked to abstain from smoking, from consuming products containing alcohol or caffeine, as well as from taking up any sort of considerable physical activity, whereas they were instructed not to fast. Room temperature was maintained constant during all measurements (- ℃). Only right eyes were tested.
... Intraocular Pressure Measurement
Non-contact tonometry was performed to assess IOP by means of a validated (Ogbuehi and Almubrad, ) device (Pulsair EasyEye, Keeler Ltd., UK). ree consecutive readings were obtained from the experimental eye and the average value was recorded.
... Retinal Functional Analysis
Details on the RVA measuring principle have been extensively described (Section ...). One drop of Tropicamide (% w/v, Bausch & Lomb, UK) was instilled to achieve pupil dilation necessary for geing unobstructed view of the posterior pole. As soon as full pupil dilation was reached, the dynamic retinal vessel assessment commenced. One retinal arteriole and one retinal venule from the superotemporal fundus area - DD away from the edge of the ONH were examined.
... Outcome Measures
e following parameters (as defined in Section ...) were obtained (from both initial and follow-up sessions) averaged across the three flicker cycles by means of raw RVA output data
Essential Hypertension: Case Reports processing: BDF, bFR, MD, MC, DA, RT, CT, ΔD and APR. Absolute arteriolar and venular diameters were recorded in MU.
.. Data Analysis
Longitudinal results are reported per individual. Due to the nature of the study (case report) no statistical analyses have been performed. For visualisation of the RVA diameter recordings, graphs were ploed comparing initial and follow-up measurements.
. Results
.. Sample
ree male Caucasians, non-smokers, essential hypertensives took part in this case reports series. JW had been previously diagnosed prior to the initial examination (treatment with combination of beta-blockers and Angiotensin Converting Enzyme (ACE) inhibitors). JH was diagnosed shortly aer the initial examination and had been under treatment (ACE inhibitors) ever since the follow-up examination. TR was undiagnosed at the time point of the initial examination and was under treatment for a few months only (ACE inhibitors and Latanoprost) prior to the follow-up visit. At the time point of their initial visit JH was , JW was and TR was years old.
.. hr BP and ECG Monitoring
Values of hour BP monitoring parameters and frequency-domain HRV parameters from hour ECG monitoring are shown in Table .. For the case of JH, sympatho-vagal balance as defined by the LF/HF ratio (for both day and night) has dropped, between the two time points, to values signifying equal sympathetic and parasympathetic activity. is was mediated from a combined drop in LF and increase in HF components. SBP and DBP values have dropped substantially, whereas HR has remained stable. Day time BP values for JW and TR have largely remained stable. Conversely, improvement on BP values due to antihypertensive medication is clearly visible in the night time values which have been lowered. Regarding frequency domain HRV parameters, the younger subject of the three (JW) shows directly opposite sympatho- vagal activity when compared to the older subject (JH) (approximately two decades of age difference). No specific paern of sympatho-vagal changes is observed in the case of TR.
.. Retinal Functional Assessment
Averaged values across the three flicker provocation cycles were calculated and reported in Table .. Both arteriolar and venular MD dropped between the two time points for JH
Essential Hypertension: Case Reports
JH, JW, TR, Parameter Initial Follow-up Initial Follow-up Initial Follow-up SBP/DBP day (mmHg) / / / / / / SBP/DBP night (mmHg) / / / / / / SBP/DBP h (mmHg) / / / / / / HR day HR night HR h LF day (NU) LF night (NU) HF day (NU) HF night (NU) LF/HF day (NU) . . . . . LF/HF night (NU) . . . . . . Day/night LF (NU) Day/night HF (NU) Day/night LF/HF (NU) . . . . . . HRV TI
Table .: BP, HR and frequency-domain HRV parameters for the initial and follow-up examinations of the three hypertensives. Noted years of age for each participant are at the time point of their initial visit. Follow-up period was five years. TI, Triangular Index; NU, Normalised Units. and JW, contrary to TR who showed increased arteriolar reactivity in both vessels. Despite the increased reactivity for the case of TR, arteriolar dilatory capacity as described by DA shows similar values across the two visits. is is corroborated by the arteriolar MC values. Apparently, antihypertensive medication across all subjects has a positive effect in arteriolar RT. Arteries take substantially less time to reach maximum dilation, although the value reached is lower for the cases of JH and JW (. MU and . MU, respectively).
Graphical representations of the diameter responses induced by flickering light by means of the RVA for the three hypertensives are shown in Figures . to .. Arteries and veins are ploed separately for easier comparisons between the initial and follow-up examinations.
. Discussion
In both hospitalised and non-hospitalised subjects electrocardiograms were recorded for hours and the main finding was that in both groups the markers of sympathetic and vagal regulation of HR underwent circadian changes (Furlan et al., ). Namely sympathetic predominance (LF component) was apparent during the day and vagal predominance (HF component) during the night. For out of instances this was true for subjects in this study as well (day LF higher than night LF and night HF higher than day HF, Table .). A more
Essential Hypertension: Case Reports
103 Initial 102 Follow-up
101
100
99 Diameter (MU) Diameter
98 Arteries - Subject JH 97 50 100 150 200 250 300 350 Time (sec)
107 Veins - Subject JH Initial 106 Follow-up 105 104 103 102 101 Diameter (MU) Diameter 100 99 98 50 100 150 200 250 300 350 Time (sec)
Figure .: Retinal vascular reactivity by means of the RVA across a year period for Subject JH. Green lines represent the follow-up measurement.
104 Initial 103 Follow-up 102 101 100 99 98 Diameter (MU) Diameter 97 96 Arteries - Subject JW 95 50 100 150 200 250 300 350 Time (sec)
106 Initial 104 Follow-up
102
100
98 Diameter (MU) Diameter
96 Veins - Subject JW 94 50 100 150 200 250 300 350 Time (sec)
Figure .: Retinal vascular reactivity by means of the RVA across a year period for Subject JW. Green lines represent the follow-up measurement.
Essential Hypertension: Case Reports
108 Arteries - Subject TR Initial 106 Follow-up 104 102 100 98 96 Diameter (MU) Diameter 94 92 90 50 100 150 200 250 300 350 Time (sec)
110 Veins - Subject TR Initial 108 Follow-up
106
104
102 Diameter (MU) Diameter
100
98 50 100 150 200 250 300 350 Time (sec)
Figure .: Retinal vascular reactivity by means of the RVA across a year period for Subject TR. Green lines represent the follow-up measurement.
JH, JW, TR, Parameter Initial Follow-up Initial Follow-up Initial Follow-up Arteries BDF (%) . (.) . (.) . (.) . (.) . () . (.) MD (%) . (.) . (.) . (.) . (.) . () . (.) MC (%) . (.) . () . (.) . (.) . () (.) DA (%) . (.) . (.) . (.) . (.) . (.) . (.) bFR (%) . (.) . (.) . () . (.) . (.) . () ΔD (%) . (.) . (.) . (.) (.) . (.) . (.) APR . (.) (.) (.) . (.) . (.) . (.) RT (s) (.) (.) (.) () (.) (.) CT (s) (.) (.) (.) (.) () (.) Veins BDF (%) . (.) (.) . (.) . (.) . (.) . (.) MD (%) . (.) . (.) . () . (.) . (.) . (.) ΔD (%) . (.) (.) . (.) . (.) . (.) . (.) RT (s) (.) (.) (.) (.) (.) (.)
Table .: RVA dynamic flicker response parameters and reaction times, averaged across all three flicker cycles. Noted years of age for each participant are at the time point of their initial visit. Follow-up period was five years. Values are expressed as means (SD). For acronyms, see page .
Essential Hypertension: Case Reports recent study reported significant alterations in markers of SA regulation (increased LF and reduced HF) both in pre-hypertensives and to a larger extent in hypertensives (Lucini et al., ). Also, they noted that “hypertensive autonomic dysregulation was particularly apparent in the youngest group”.
As it can be seen from the limited amount of data of this small case report study, the interactions between factors, such as, disease onset and duration, treatment type and duration, age and others are creating a highly diverse clinical picture. An additional limitation of the case reports presented here is the long follow-up period of years with no in between repeat visits. Although data shown here include a temporal element, more frequent examinations should ideally be performed (e.g. on a yearly basis). Similar studies to the concept of Nagel et al. (b) but with sufficiently large sample sizes and well stratified hypertensive groups would enable fruitful comparisons across varying pathological and physiological states. Nevertheless, the combination of macrovascular with microvascular information is the necessary step for the ultimate goal for risk stratification or treatment monitoring.
Each year CVD causes more than . million deaths in Europe, accounting for nearly half of all deaths (%) (Allender et al., ). Major advances in prevention have led to improved hypertension-related mortality and morbidity figures over the last three decades; nevertheless essential hypertension is still the most prevalent among cardiovascular disorders. e pathophysiology of essential hypertension involves a multitude of factors, including the central nervous system, endocrine factors, the large arteries and the microcirculation. Large scale population-based studies pioneered in identifying the relationships between impaired microvascular perfusion, autoregulation or structure and subsequent target organ damage. e limitation of such studies is that they are cross-sectional in nature. is allows only assumptions to be made on the time course of the various disease manifestations: does microvascular damage precede CVD or vice versa, or is it a complex two-way interaction?
No single modality is able to give definitive answers. e techniques as well as the strategies for investigating microcirculatory function have evolved almost exponentially over the last years. e RVA technology is definitely a piece of the puzzle. It may serve as a complementary research tool to help answer unresolved questions in both physiologic and pathological conditions. Existing data demonstrate that visual stimulation is a powerful modulator of retinal and optic nerve blood flow. Much work has been accomplished thus far in exploring neurovascular/neurometabolic coupling in the human retina but there are still many open questions that remain to be elucidated. Technical limitations in imaging methods of low spatial and temporal resolution can now be overcome making the reliable tracking of retinal capillaries feasible (Bedggood and Metha, ).
Bibliography
Abramoff, M., Magalhaes, P., and Ram, S. (). Image processing with ImageJ. Biophotonics international, ():–.
Allender, S., Scarborough, P., Peto, V., Rayner, M., Leal, J., Luengo-Fernandez, R., and Gray, A. (). European cardiovascular disease statistics. European Heart Network London.
Antonios, T. F. T., Singer, D. R. J., Markandu, N. D., Mortimer, P. S., and MacGregor, G. A. (). Structural Skin Capillary Rarefaction in Essential Hypertension. Hypertension, ():– .
Anuradha, S., Healy, G. N., Dunstan, D. W., Klein, R., Klein, B. E., Cotch, M. F., Wong, T. Y., and Owen, N. (). Physical Activity, Television Viewing Time, and Retinal Microvascular Caliber. American Journal of Epidemiology, ():–.
Azegrouz, H., Trucco, E., Dhillon, B., MacGillivray, T., and MacCormick, I. (). ickness dependent tortuosity estimation for retinal blood vessels. In Engineering in Medicine and Biology Society, . EMBS ’. th Annual International Conference of the IEEE, volume , pages – .
Barna, I., Keszei, A., and Dunai, A. (). Evaluation of Meditech ABPM- ambulatory blood pressure measuring device according to the British Hypertension Society protocol. Blood Press Monitoring, ():–.
Barold, S. (). Norman J. “Je” Holter–“Father” of Ambulatory ECG Monitoring. Journal of Interventional Cardiac Electrophysiology, ():–.
Bedggood, P. and Metha, A. (). Direct visualization and characterization of erythrocyte flow in human retinal capillaries. Biomed. Opt. Express, ():–.
Beevers, G., Lip, G. Y., and O’Brien, E. (). Blood pressure measurement: Part I— Sphygmomanometry: factors common to all techniques. BMJ, ():–.
Bek, T., Hajari, J., and Jeppesen, P. (). Interaction between flicker-induced vasodilatation and pressure autoregulation in early retinopathy of type diabetes. Graefe’s Archive for
Bibliography
Clinical and Experimental Ophthalmology, ():–.
Berne, R. M. and Levy, M. N. (). Cardiovascular Physiology. Mosby, th edition.
Bevan, J. A., Bevan, R. D., Hwa, J. J., Owen, M. P., and Tayo, F. M. (). Calcium Regulation in Vascular Smooth Muscle: Is ere a Paern to Its Variability Within the Arterial Tree? Journal of Cardiovascular Pharmacology, :S–S.
Blum, M., Bachmann, K., Wintzer, D., Riemer, T., Vilser, W., and Strobel, J. (). Noninvasive measurement of the Bayliss effect in retinal autoregulation. Graefe’s Archive for Clinical and Experimental Ophthalmology, ():–.
Blum, M., Kloos, C., Günther, S., Hunger-Dathe, W., and Müller, U. (). Improved metabolic control results in beer myogenic response of retinal arterioles in patients with diabetes mellitus type . Ophthalmologica, ():–.
Blum, M., Scherf, C., Bachmann, K., and Strobel, J. (). Alterskorrelierte Kontraktilität retinaler Arteriolen bei Sauerstoffatmung. Der Ophthalmologe, :–.
Bracher, D. (). Changes in peripapillary tortuosity of the central retinal arteries in newborns. Graefe’s Archive for Clinical and Experimental Ophthalmology, :–.
Brown, S. M. and Jampol, L. M. (). New Concepts of Regulation of Retinal Vessel Tone. Arch Ophthalmol, ():–.
Chapman, N., Dell’omo, G., Sartini, M. S., Wi, N., Hughes, A., om, S., and Pedrinelli, R. (). Peripheral vascular disease is associated with abnormal arteriolar diameter relationships at bifurcations in the human retina. Clin. Sci., ():–.
Chapman, N., Haimes, G., Stanton, A. V., om, S. A., and Hughes, A. D. (). Acute effects of oxygen and carbon dioxide on retinal vascular network geometry in hypertensive and normotensive subjects. Clin. Sci., ():–.
Chasis, H. (). Appreciation of the Keith, Wagener, and Barker classification of hypertensive disease. e American Journal of the Medical Sciences, ():.
Chen, H., Patel, V., Wiek, J., Rassam, S., and Kohner, E. (). Vessel diameter changes during the cardiac cycle. Eye (London, England), :–.
Cheung, C. Y., Hsu, W., Lee, M. L., Wang, J. J., Mitchell, P., Lau, Q. P., Hamzah, H., Ho, M., and Wong, T. Y. (). A New Method to Measure Peripheral Retinal Vascular Caliber over an Extended Area. Microcirculation, ():–.
Clifford, G. D., Azuaje, F., and McSharry, P. (). Advanced Methods And Tools for ECG Data Analysis. Artech House, Inc.
Bibliography
Dart, A. M. and Kingwell, B. A. (). Pulse pressure—a review of mechanisms and clinical relevance. Journal of the American College of Cardiology, ():–.
Dawczynski, J., Mandecka, A., Blum, M., Müller, U. A., Ach, T., and Strobel, J. (). Endotheliale Dysfunktion zentraler Netzhautgeäße als Prognoseparameter der diabetischen Retinopathie? Klin Monatsbl Augenheilkd, ():–.
Delles, C., Michelson, G., Harazny, J., Oehmer, S., Hilgers, K. F., and Schmieder, R. E. (). Impaired Endothelial Function of the Retinal Vasculature in Hypertensive Patients. Stroke, ():–.
Dolan, E., Li, Y., ijs, L., McCormack, P., Staessen, J. A., O’Brien, E., and Stanton, A. (). Ambulatory arterial stiffness index: rationale and methodology. Blood Pressure Monitoring, ():–.
Dorion, T. (). Manual of ocular fundus examination. Buerworth-Heinemann.
Dougherty, G., Johnson, M., and Wiers, M. (). Measurement of retinal vascular tortuosity and its application to retinal pathologies. Medical and Biological Engineering and Computing, ():–.
Dougherty, G. and Varro, J. (). A quantitative index for the measurement of the tortuosity of blood vessels. Medical Engineering & Physics, (): – .
Dumskyj, M. J., Aldington, S. J., Doré, C. J., and Kohner, E. M. (). e accurate assessment of changes in retinal vessel diameter using multiple frame electrocardiograph synchronised fundus photography. Current Eye Research, ():–.
Faul, F., Erdfelder, E., Lang, A.-G., and Buchner, A. (). G*power : a flexible statistical power analysis program for the social, behavioral, and biomedical sciences. Behav Res Methods, ():–.
Formaz, F., Riva, C., and Geiser, M. (). Diffuse luminance flicker increases retinal vessel diameter in humans. Current Eye Research, ():–.
Frederiksen, C., Jeppesen, P., Knudsen, S., Poulsen, P., Mogensen, C., and Bek, T. (). e blood pressure-induced diameter response of retinal arterioles decreases with increasing diabetic maculopathy. Graefe’s Archive for Clinical and Experimental Ophthalmology, ():–.
Furlan, R., Guzzei, S., Crivellaro, W., Dassi, S., Tinelli, M., Baselli, G., Cerui, S., Lombardi, F., Pagani, M., and Malliani, A. (). Continuous -hour assessment of the neural regulation of systemic arterial pressure and RR variabilities in ambulant subjects. Circulation, ():– .
Bibliography
Gafiychuk, V. V. and Lubashevsky, I. A. (). On the Principles of the Vascular Network Branching. Journal of eoretical Biology, (): – .
Garhöfer, G., Bek, T., Boehm, A. G., Gherghel, D., Grunwald, J., Jeppesen, P., Kergoat, H., Kotliar, K., Lanzl, I., Lovasik, J. V., Nagel, E., Vilser, W., Orgul, S., and Schmeerer, L. (). Use of the retinal vessel analyzer in ocular blood flow research. Acta Ophthalmologica, ():–.
Garhöfer, G., Resch, H., Weigert, G., Lung, S., Simader, C., and Schmeerer, L. (). Short- Term Increase of Intraocular Pressure Does Not Alter the Response of Retinal and Optic Nerve Head Blood Flow to Flicker Stimulation. Investigative Ophthalmology & Visual Science, ():–.
Garhöfer, G., Zawinka, C., Huemer, K.-H., Schmeerer, L., and Dorner, G. T. (). Flicker Light–Induced Vasodilatation in the Human Retina: Effect of Lactate and Changes in Mean Arterial Pressure. Investigative Ophthalmology & Visual Science, ():–.
Gasser, P. and Bühler, F. R. (). Nailfold microcirculation in normotensive and essential hypertensive subjects, as assessed by video-microscopy. Journal of Hypertension, ():– .
Ghiadoni, L., Faita, F., Salvei, M., Cordiano, C., Biggi, A., Puato, M., Di Monaco, A., De Siati, L., Volpe, M., Ambrosio, G., Gemignani, V., Muiesan, M. L., Taddei, S., Lanza, G. A., and Cosentino, F. (). Assessment of flow-mediated dilation reproducibility: a nationwide multicenter study. Journal of Hypertension, ():–.
Gugleta, K., Kochkorov, A., Katamay, R., Zawinka, C., Flammer, J., and Orgul, S. (). Analysis of Retinal Vasodilation aer Flicker Light Stimulation in Relation to Vasospastic Propensity. Investigative Ophthalmology & Visual Science, ():–.
Gugleta, K., Kochkorov, A., Waldmann, N., Polunina, A., Katamay, R., Flammer, J., and Orgul, S. (). Dynamics of retinal vessel response to flicker light in glaucoma patients and ocular hypertensives. Graefe’s Archive for Clinical and Experimental Ophthalmology, ():–.
Gugleta, K., Waldmann, N., Polunina, A., Kochkorov, A., Katamay, R., Flammer, J., and Orgul, S. (). Retinal neurovascular coupling in patients with glaucoma and ocular hypertension and its association with the level of glaucomatous damage. Graefe’s Archive for Clinical and Experimental Ophthalmology, :–.
Gunn, M. (). On ophthalmoscopic evidence of general arterial disease. Trans Ophthalmol Soc UK, :–.
Guzzei, S., Dassi, S., Pecis, M., Casat, R., Masu, A. M., Longoni, P., Tinelli, M., Cerui, S., Pagani, M., and Malliani, A. (). Altered paern of circadian neural control of heart
Bibliography
period in mild hypertension. Journal of Hypertension, ():–.
Haefliger, I., Flammer, J., and Luscher, T. (). Heterogeneity of endothelium-dependent regulation in ophthalmic and ciliary arteries. Investigative Ophthalmology & Visual Science, ():–.
Hammer, M., Vilser, W., Riemer, T., and Schweitzer, D. (). Retinal vessel oximetry- calibration, compensation for vessel diameter and fundus pigmentation, and reproducibility. Journal of Biomedical Optics, ():–––.
Harper, R. N., Moore, M. A., Marr, M. C., Was, L. E., and Hutchins, P. M. (). Arteriolar rarefaction in the conjunctiva of human essential hypertensives. Microvascular Research, ():–.
Hart, W. E., Goldbaum, M., Côté, B., Kube, P., and Nelson, M. R. (). Measurement and classification of retinal vascular tortuosity. International Journal of Medical Informatics, (- ): – .
Heitmar, R., Blann, A., Cubbidge, R. P., Lip, G., and Gherghel, D. (). Continuous retinal vessel diameter measurements - the future of retinal vessel assessment? Investigative Ophthalmology & Visual Science, ():–.
Heitmar, R., Blann, A., Cubbidge, R. P., Lip, G., and Gherghel, D. (a). Author Response: Can Vascular Function Be Assessed by the Interpretation of Retinal Vascular Diameter Changes? Investigative Ophthalmology & Visual Science, ():–.
Heitmar, R., Cubbidge, R. P., Lip, G. Y. H., Gherghel, D., and Blann, A. D. (b). Altered Blood Vessel Responses in the Eye and Finger in Coronary Artery Disease. Investigative Ophthalmology & Visual Science, ():–.
Heitmar, R. and Summers, R. (). Assessing vascular function using dynamic retinal diameter measurements: A new insight on the endothelium. rombosis and Haemostasis, ():–.
Hemminki, V., Kähönen, M., Tuomisto, M., Turjanmaa, V., and Uusitalo, H. (). Determination of retinal blood vessel diameters and arteriovenous ratios in systemic hypertension: comparison of different calculation formulae. Graefe’s Archive for Clinical and Experimental Ophthalmology, :–.
Hnatkova, K., Copie, X., Staunton, A., and Malik, M. (). Numeric processing of Lorenz plots of R-R intervals from long-term ECGs: Comparison with time-domain measures of heart rate variability for risk stratification aer myocardial infarction. Journal of Electrocardiology, , Supplement (): – .
Bibliography
Houben, A., Canoy, M., Paling, H., Derhaag, P., and de Leeuw, P. (). antitative analysis of retinal vascular changes in essential and renovascular hypertension. J Hypertens, ():–.
Hubbard, L., Ehrhardt, B., and Klein, R. (). e association between generalized arteriolar narrowing and blood pressure. Investigative Ophthalmology & Visual Science, :.
Hubbard, L. D., Brothers, R. J., King, W. N., Clegg, L. X., Klein, R., Cooper, L. S., Sharre, A. R., Davis, M. D., and Cai, J. (). Methods for evaluation of retinal microvascular abnormalities associated with hypertension/sclerosis in the Atherosclerosis Risk In Communities study. Ophthalmology, (): – .
Hughes, A. D., Martinez-Perez, E., Jabbar, A. S., Hassan, A., Wi, N. W., Mistry, P. D., Chapman, N., Stanton, A. V., Beevers, G., Pedrinelli, R., Parker, K. H., and om, S. A. (). antification of topological changes in retinal vascular architecture in essential and malignant hypertension. J Hypertens, ():–.
Jacks, A. S. and Miller, N. R. (). Spontaneous retinal venous pulsation: aetiology and significance. Journal of Neurology, Neurosurgery & Psychiatry, ():–.
Jean-Louis, S., Lovasik, J., and Kergoat, H. (). Systemic hyperoxia and retinal vasomotor responses. Investigative Ophthalmology & Visual Science, ():–.
Jensen, S. P., Jeppesen, P., and Bek, T. (). Differential diameter responses in macular and peripheral retinal arterioles may contribute to the regional distribution of diabetic retinopathy lesions. Graefe’s Archive for Clinical and Experimental Ophthalmology, :– .
Jeppesen, P., Sanye-Hajari, J., and Bek, T. (). Increased Blood Pressure Induces a Diameter Response of Retinal Arterioles that Increases with Decreasing Arteriolar Diameter. Investigative Ophthalmology & Visual Science, ():–.
Johnson, P. (). Autoregulation of blood flow. Circulation research, ():–.
Kagan, A., Aurell, E., and Tibblin, G. (). Signs in the fundus oculi and arterial hypertension: unconventional assessment and significance. Bulletin of the World Health Organization, ():–.
Kalitzeos, A. A., Lip, G. Y., and Heitmar, R. (). Retinal vessel tortuosity measures and their applications. Experimental Eye Research, (): – .
Kamath, M. and Fallen, E. (). Power spectral analysis of heart rate variability: a noninvasive signature of cardiac autonomic function. Crit Rev Biomed Eng, ():–.
Bibliography
Keith, N. M. M., Wagener, H. P. M., and Barker, N. W. M. (). Some different types of essential hypertension: their course and prognosis. e American Journal of the Medical Sciences, ():–.
King, L., Stanton, A., Sever, P., om, S., and Hughes, A. (). Arteriolar length-diameter (L:D) ratio: a geometric parameter of the retinal vasculature diagnostic of hypertension. Journal of Human Hypertension, ():–.
Kiss, B., Polska, E., Dorner, G., Polak, K., Findl, O., Mayrl, G. F., Eichler, H.-G., Wolzt, M., and Schmeerer, L. (). Retinal Blood Flow during Hyperoxia in Humans Revisited: Concerted Results Using Different Measurement Techniques. Microvascular Research, (): – .
Klabunde, R. E. (). Cardiovascular physiology concepts. Wolters Kluwer Health.
Knudtson, M. D., Klein, B. E. K., Klein, R., Wong, T. Y., Hubbard, L. D., Lee, K. E., Meuer, S. M., and Bulla, C. P. (). Variation associated with measurement of retinal vessel diameters at different points in the pulse cycle. British Journal of Ophthalmology, ():–.
Knudtson, M. D., Lee, K. E., Hubbard, L. D., Wong, T. Y., Klein, R., and Klein, B. E. (). Revised formulas for summarizing retinal vessel diameters. Current Eye Research, ():–.
Kotliar, K., Nagel, E., Vilser, W., and Lanzl, I. (). Functional in vivo assessment of retinal artery microirregularities in glaucoma. Acta Ophthalmologica, ():–.
Kotliar, K., Nagel, E., Vilser, W., Seidova, S.-F., and Lanzl, I. M. (). Microstructural alterations of retinal arterial blood column along the vessel axis in systemic hypertension. Investigative Ophthalmology & Visual Science, ():–.
Kotliar, K. E., Baumann, M., Vilser, W., and Lanzl, I. M. (a). Pulse wave velocity in retinal arteries of healthy volunteers. British Journal of Ophthalmology, :–.
Kotliar, K. E., Lanzl, I. M., Schmidt-Trucksäss, A., Sitnikova, D., Ali, M., Blume, K., Halle, M., and Hanssen, H. (b). Dynamic retinal vessel response to flicker in obesity: A methodological approach. Microvascular Research, (): – .
Kotliar, K. E., Vilser, W., Nagel, E., and Lanzl, I. (). Retinal vessel reaction in response to chromatic flickering light. Graefe’s Archive for Clinical and Experimental Ophthalmology, :–.
Kylstra, J., Wierzbicki, T., Wolbarsht, M., Landers, M., and Stefansson, E. (). e relationship between retinal vessel tortuosity, diameter, and transmural pressure. Graefe’s Archive for Clinical and Experimental Ophthalmology, :–.
Bibliography
La Rovere, M. T., Specchia, G., Mortara, A., and Schwartz, P. J. (). Baroreflex sensitivity, clinical correlates, and cardiovascular mortality among patients with a first myocardial infarction. A prospective study. Circulation, ():–.
Langewitz, W., Rüddel, H., and Schächinger, H. (). Reduced parasympathetic cardiac control in patients with hypertension at rest and under mental stress. American Heart Journal, (): – .
Lanigan, L., Clark, C., and Hill, D. (). Retinal circulation responses to systemic autonomic nerve stimulation. Eye (London, England), :–.
Lanzl, I., Seidova, S., Maier, M., Lohmann, C., Schmidt-Trucksäss, A., Halle, M., and Kotliar, K. (). Dynamic retinal vessel response to flicker in age-related macular degeneration patients before and aer vascular endothelial growth factor inhibitor injection. Acta Ophthalmologica, :–.
Lasta, M., Palkovits, S., Boltz, A., Schmidl, D., Kaya, S., Cherecheanu, A. P., Garhöfer, G., and Schmeerer, L. (). Reproducibility of retinal vessel oxygen saturation measurements in healthy young subjects. Acta Ophthalmologica, :e–e.
Lasta, M., Pemp, B., Schmidl, D., Boltz, A., Kaya, S., Palkovits, S., Werkmeister, R., Howorka, K., Popa-Cherecheanu, A., Garhöfer, G., and Schmeerer, L. (). Neurovascular Dysfunction Precedes Neural Dysfunction in the Retina of Patients with Type Diabetes. Investigative Ophthalmology & Visual Science, ():–.
Leatham, A. (). e retinal vessels in hypertension. QJM, ():–.
Lee, B. B., Pokorny, J., Smith, V. C., Martin, P. R., and Valberg, A. (). Luminance and chromatic modulation sensitivity of macaque ganglion cells and human observers. JOSA A, ():–.
Li, Y., Wang, J.-G., Dolan, E., Gao, P.-J., Guo, H.-F., Nawrot, T., Stanton, A. V., Zhu, D.-L., O’Brien, E., and Staessen, J. A. (). Ambulatory Arterial Stiffness Index Derived From -Hour Ambulatory Blood Pressure Monitoring. Hypertension, ():–.
Lombardi, F., Malliani, A., Pagani, M., and Cerui, S. (). Heart rate variability and its sympatho-vagal modulation. Cardiovascular research, ():–.
Lotmar, W., Freiburghaus, A., and Bracher, D. (). Measurement of vessel tortuosity on fundus photographs. Graefe’s Archive for Clinical and Experimental Ophthalmology, :– .
Lo, M. E. J., Slocomb, J. E., Shivkumar, V., Smith, B., Gabbay, R. A., illen, D., Gardner, T. W., and Beermann, K. (). Comparison of retinal vasodilator and constrictor responses in
Bibliography
type diabetes. Acta Ophthalmologica, ():e–e.
Lucini, D., Mela, G. S., Malliani, A., and Pagani, M. (). Impairment in cardiac autonomic regulation preceding arterial hypertension in humans: Insights from spectral analysis of beat-by-beat cardiovascular variability. Circulation, ():–.
Luscher, T. F. (). e Endothelium and Cardiovascular Disease – A Complex Relation. N Engl J Med, ():–.
Mahler, F., Saner, H., Boss, C., and Annaheim, M. (). Local cold exposure test for capillaroscopic examination of patients with Raynaud’s syndrome. Microvascular research, ():–.
Malik, M., Bigger, J. T., Camm, A. J., Kleiger, R. E., Malliani, A., Moss, A. J., and Schwartz, P. J. (). Heart rate variability: Standards of measurement, physiological interpretation, and clinical use. European Heart Journal, ():–.
Malliani, A. (). Heart rate variability: from bench to bedside . European Journal of Internal Medicine, (): – .
Man, R. E. K., Kawasaki, R., Wu, Z., Luu, C. D., Wang, J. J., Wong, T. Y., and Lamoureux, E. L. (). Reliability and Reproducibility of Retinal Oxygen Saturation Measurements using a Predefined Peri-papillary Annulus. Acta Ophthalmologica, :e–e.
Mancia, G., De Backer, G., Dominiczak, A., Ciova, R., Fagard, R., Germano, G., Grassi, G., Heagerty, A. M., Kjeldse n, S. E., Laurent, S., Narkiewicz, K., Ruilope, L., Rynkiewicz, A., Schmieder, R. E., Struijker Boudier, H. A., Zanchei, A. t., Vahanian, A., Camm, J., De Caterina, R., Dean, V., Dickstein, K., Filippatos, G., Funck-Brentano, C., Hellemans, I., Kristensen, S. D., McGregor, K., Sechtem, U., Silber, S., Tendera, M., Widimsky, P., Zamorano, J. L., Kjeldsen, S. E., Erdine, S., Narkiewicz, K., Kiowski, W., Agabiti-Rosei, E., Ambrosioni, E., Ciova, R., Dominiczak, A., Fagard, R., Heagerty, A. h. M., Laurent, S., Lindholm, L. H., Mancia, G., Manolis, A., Nilsson, P. M., Redon, J., Schmieder, R. E., Struijker-Boudier, H. A., Viigimaa, M., Filippatos, G., Adamopoulos, S., Agabiti-Rosei, E., Ambrosioni, E., Bertomeu, V., Clement, D., Er dine, S., Farsang, C., Gaita, D., Kiowski, W., Lip, G., Mallion, J.-M., Manolis, A. J., Nilsson, P. M., O’Brien, E., P onikowski, P., Redon, J., Ruschitzka, F., Tamargo, J., van Zwieten, P., Viigimaa, M., Waeber, B., Williams, B., and Zamorano, J. L. (). Guidelines for the management of arterial hypertension. European Heart Journal, ():–.
Mandecka, A., Dawczynski, J., Blum, M., Müller, N., Kloos, C., Wolf, G., Vilser, W., Hoyer, H., and Müller, U. A. (). Influence of Flickering Light on the Retinal Vessels in Diabetic Patients. Diabetes Care, ():–.
Bibliography
Mandecka, A., Dawczynski, J., Vilser, W., Blum, M., Müller, N., Kloos, C., Wolf, G., and Müller, U. A. (). Abnormal retinal autoregulation is detected by provoked stimulation with flicker light in well-controlled patients with type diabetes without retinopathy. Diabetes Research and Clinical Practice, (): – .
Mehlsen, J., Jeppesen, P., Erlandsen, M., Poulsen, P. L., and Bek, T. (). Lack of effect of short- term treatment with Amlodipine and Lisinopril on retinal autoregulation in normotensive patients with type diabetes and mild diabetic retinopathy. Acta Ophthalmologica, :– .
Millar-Craig, M., Bishop, C., and Raery, E. (). Circadian variation of blood pressure. e Lancet, (): – .
Moore, R. F. (). e Retinitis of Arterio-Sclerosis, and Its Relation to Renal Retinitis and to Cerebral Vascular Disease. QJM, os-(-):–.
Muench, K., Vilser, W., and Senff, I. (). Adaptive Algorithms for the Automatic Measurement of Retinal Vessel Diameters. Biomedizinische Technik, ():–.
Murray, C. (). e physiological principle of minimum work applied to the angle of branching of arteries. Proc Natl Acad Sci U S A, ():–.
Nadar, S., Blann, A. D., and Lip, G. Y. (). Endothelial Dysfunction: Methods of Assessment and Application to Hypertension. Current Pharmaceutical Design, ():–.
Nagel, E. and Vilser, W. (). Autoregulative behavior of retinal arteries and veins during changes of perfusion pressure: a clinical study. Graefe’s Archive for Clinical and Experimental Ophthalmology, :–.
Nagel, E., Vilser, W., Fink, A., and Riemer, T. (a). Varianz der Netzhautgeäßreaktion auf Flickerlicht. Der Ophthalmologe, ():–.
Nagel, E., Vilser, W., Fink, A., Riemer, T., and Lanzl, I. (b). Blood pressure effects on retinal vessel diameter and flicker response: A .-year follow-up. European Journal of Ophthalmology, ():–.
Nagel, E., Vilser, W., and Lanzl, I. (). Age, Blood Pressure, and Vessel Diameter as Factors Influencing the Arterial Retinal Flicker Response. Investigative Ophthalmology & Visual Science, ():–.
Nagel, E., Vilser, W., and Lanzl, I. (). Vergleich der Durchmesserreaktion retinaler Arterien und Venen auf Flickerlicht. Der Ophthalmologe, ():–.
Bibliography
Nguyen, T. T., Kreis, A. J., Kawasaki, R., Wang, J. J., Seifert, B.-U., Vilser, W., Nagel, E., and Wong, T. Y. (). Reproducibility of the Retinal Vascular Response to Flicker Light in Asians. Current Eye Research, ():–.
Nickla, D. L. and Wallman, J. (). e multifunctional choroid. Progress in Retinal and Eye Research, (): – .
Noon, J. P., Walker, B. R., Webb, D. J., Shore, A. C., Holton, D. W., Edwards, H. V., and Wa, G. C. (). Impaired microvascular dilatation and capillary rarefaction in young adults with a predisposition to high blood pressure. J Clin Invest, ():–.
Ogbuehi, K. C. and Almubrad, T. M. (). Accuracy and Reliability of the Keeler Pulsair EasyEye Non-Contact Tonometer. Optometry & Vision Science, ():–.
Pache, M., Nagel, E., and Flammer, J. (). Reproduzierbarkeit der Messungen mit dem Retinal Vessel Analyzer unter Optimalbedingungen Reproducibility of measurements with the retinal vessel analyzer under optimal conditions. Klinische Monatsbläer ür Augenheilkunde, :–.
Pagani, M., Lombardi, F., Guzzei, S., Rimoldi, O., Furlan, R., Pizzinelli, P., Sandrone, G., Malfao, G., Dell’Orto, S., and Piccaluga, E. (). Power spectral analysis of heart rate and arterial pressure variabilities as a marker of sympatho-vagal interaction in man and conscious dog. Circulation Research, ():–.
Panza, J. A., yyumi, A. A., Brush, J. E., and Epstein, S. E. (). Abnormal Endothelium- Dependent Vascular Relaxation in Patients with Essential Hypertension. New England Journal of Medicine, ():–.
Parr, J. (). Hypertensive generalised narrowing of retinal arteries. Transactions of the Ophthalmological Society of New Zealand, ():–.
Parr, J. and Spears, G. (a). General caliber of the retinal arteries expressed as the equivalent width of the central retinal artery. American journal of ophthalmology, ():–.
Parr, J. and Spears, G. (b). Mathematic relationships between the width of a retinal artery and the widths of its branches. American journal of ophthalmology, ():–.
Patel, S. R., Bellary, S., Qin, L., Gill, P., Taheri, S., Heitmar, R., Gibson, J. M., and Gherghel, D. (). Abnormal retinal vascular function and lipid levels in a sample of healthy UK South Asians. British Journal of Ophthalmology, :–.
Paon, N., Aslam, T., MacGillivray, T., Dhillon, B., and Constable, I. (). Asymmetry of Retinal Arteriolar Branch Widths at Junctions Affects Ability of Formulæ to Predict Trunk Arteriolar Widths. Investigative Ophthalmology & Visual Science, ():–.
Bibliography
Paon, N., Maini, R., MacGillivary, T., Aslam, T. M., Deary, I. J., and Dhillon, B. (). Effect of Axial Length on Retinal Vascular Network Geometry. American Journal of Ophthalmology, ():.e – .e.
Pemp, B., Weigert, G., Karl, K., Petzl, U., Wolzt, M., Schmeerer, L., and Garhöfer, G. (). Correlation of Flicker-Induced and Flow-Mediated Vasodilatation in Patients With Endothelial Dysfunction and Healthy Volunteers. Diabetes Care, ():–.
Pickering, T. G., Gerin, W., and Schwartz, A. R. (). What is the white-coat effect and how should it be measured? Blood pressure monitoring, ():–.
Polak, K., Dorner, G., Kiss, B., Polska, E., Findl, O., Rainer, G., Eichler, H.-G., and Schmeerer, L. (). Evaluation of the Zeiss retinal vessel analyser. British Journal of Ophthalmology, ():–.
Polak, K., Schmeerer, L., and Riva, C. E. (). Influence of Flicker Frequency on Flicker- Induced Changes of Retinal Vessel Diameter. Investigative Ophthalmology & Visual Science, ():–.
Pournaras, C. J., Rungger-Brändle, E., Riva, C. E., Hardarson, S. H., and Stefansson, E. (). Regulation of retinal blood flow in health and disease. Progress in Retinal and Eye Research, (): – .
yyumi, A. A., Crake, T., Mockus, L. J., Wright, C. A., Rickards, A. F., and Fox, K. M. (). Value of the bipolar lead CM in electrocardiography. British Heart Journal, ():–.
Reimann, M., Prieur, S., Lippold, B., Bornstein, S. R., Reichmann, H., Julius, U., and Ziemssen, T. (). Retinal vessel analysis in hypercholesterolemic patients before and aer LDL apheresis. Atherosclerosis Supplements, (): – . Proceedings from the st Dresden International Symposium on erapeutic Apheresis: Recent progress in erapeutic Apheresis.
Resch, H., Zawinka, C., Weigert, G., Schmeerer, L., and Garhöfer, G. (). Inhaled Carbon Monoxide Increases Retinal and Choroidal Blood Flow in Healthy Humans. Investigative Ophthalmology & Visual Science, ():–.
Rickenbacher, I., Gugleta, K., Zawinka, C., Schötzau, A., Katamay, R., Flammer, J., and Orgül, S. (). Flickerlichtprovokation bei Vasospastikern verglichen mit gesunden Kontrollpersonen. Klinische Monatsbläer ür Augenheilkunde, :–.
Riva, C. E., Logean, E., and Falsini, B. (). Visually evoked hemodynamical response and assessment of neurovascular coupling in the optic nerve and retina. Progress in Retinal and Eye Research, (): – .
Bibliography
Rogers, K. (). Blood: Physiology and Circulation. e Rosen Publishing Group.
Rogoza, A. N., Pavlova, T. S., and Sergeeva, M. V. (). Validation of A&D UA- device for the self-measurement of blood pressure. Blood Pressure Monitoring, ():–.
Rueddel, T., Kneser, M., and Tost, F. (). Impact of exercise on retinal microvascular regulation measured by dynamic vessel analysis in healthy individuals. Clinical Physiology and Functional Imaging, ():–.
Scheie, H. (). Evaluation of Ophthalmoscopic changes of Hypertension and Arteriolar Sclerosis. AMA Arch Ophthalmol, ():–.
Seifertl, B. U. and Vilser, W. (). Retinal Vessel Analyzer (RVA)–design and function. Biomed Tech, Suppl Pt :–.
Serne, E. H., Gans, R. O., ter Maaten, J. C., Tangelder, G.-J., Donker, A. J., and Stehouwer, C. D. (). Impaired Skin Capillary Recruitment in Essential Hypertension Is Caused by Both Functional and Structural Capillary Rarefaction. Hypertension, ():–.
Shantsila, A., Shantsila, E., and YH Lip, G. (). Malignant Hypertension: A Rare Problem or is it Underdiagnosed? Current Vascular Pharmacology, ():–.
Shrout, P. and Fleiss, J. (). Intraclass correlations: uses in assessing rater reliability. Psychological bulletin, ():–.
Staessen, J. A., Wang, J., Bianchi, G., and Birkenhäger, W. H. (). Essential hypertension. e Lancet, (): – .
Stanton, A., Wasan, B., Cerui, A., Ford, S., Marsh, R., Sever, P., om, S., and Hughes, A. (). Vascular network changes in the retina with age and hypertension. J Hypertens, ( Pt ):–.
Stokoe, N. and Turner, R. (). Normal retinal vascular paern. Arteriovenous ratio as a measure of arterial calibre. British Medical Journal, ():–.
Stücker, M., Baier, V., Reuther, T., Hoffmann, K., Kellam, K., and Altmeyer, P. (). Capillary blood cell velocity in human skin capillaries located perpendicularly to the skin surface: measured by a new laser Doppler anemometer. Microvascular research, :–.
Taarnhøj, N. C. B. B., Larsen, M., Sander, B., Kyvik, K. O., Kessel, L., Hougaard, J. L., and Sorensen, T. I. A. (). Heritability of Retinal Vessel Diameters and Blood Pressure: A Twin Study. Investigative Ophthalmology & Visual Science, ():–.
Taarnhøj, N. C. B. B., Munch, I. C., Sander, B., Kessel, L., Hougaard, J. L., Kyvik, K., Sørensen, T. I. A., and Larsen, M. (). Straight versus tortuous retinal arteries in relation to blood
Bibliography
pressure and genetics. British Journal of Ophthalmology, ():–.
Taylor, J. A., Carr, D. L., Myers, C. W., and Eckberg, D. L. (). Mechanisms Underlying Very-Low-Frequency RR-Interval Oscillations in Humans. Circulation, ():–.
Verberk, W., Kroon, A., Jongen-Vancraybex, H., and De Leeuw, P. (). e applicability of home blood pressure measurement in clinical practice: A review of literature. Vascular Health and Risk Management, ():–.
Verdecchia, P. (). Prognostic Value of Ambulatory Blood Pressure: Current Evidence and Clinical Implications. Hypertension, ():–.
Waeber, B., Genoud, M., Feihl, F., Hayoz, D., and Waeber, G. (). Ambulatory blood pressure monitoring: a mean to stratify cardiovascular risk. Blood Pressure Monitoring, ():–.
Wallace, D. (). Computer-assisted quantification of vascular tortuosity in retinopathy of prematurity (an American Ophthalmological Society thesis). Transactions of the American Ophthalmological Society, :–.
Walsh, J. (). Hypertensive retinopathy. Description, classification, and prognosis. Ophthalmology, ():–.
White, W. B., Berson, A. S., Robbins, C., Jamieson, M. J., Prisant, L. M., Roccella, E., and Sheps, S. G. (). National standard for measurement of resting and ambulatory blood pressures with automated sphygmomanometers. Hypertension, ():–.
Whitworth, J. (). World Health Organization (WHO)/International Society of Hypertension (ISH) statement on management of hypertension. Journal of hypertension, ():–.
Williams, B., Poulter, N., Brown, M., Davis, M., McInnes, G., Poer, J., Sever, P., and om, S. (). Guidelines for management of hypertension: report of the fourth working party of the British Hypertension Society, —BHS IV. Journal of Human Hypertension, ():– .
Williams, T. (). antification of arteriolar tortuosity in two normotensive age groups. American journal of optometry and physiological optics, ():–.
Wimpissinger, B., Resch, H., Berisha, F., Weigert, G., Schmeerer, L., and Polak, K. (). Response of choroidal blood flow to carbogen breathing in smokers and non-smokers. British Journal of Ophthalmology, ():–.
Wi, N. W., Chapman, N., om, S. A. M., Stanton, A. V., Parker, K. H., and Hughes, A. D. (). A novel measure to characterise optimality of diameter relationships at retinal vascular
Bibliography
bifurcations. Artery Research, (): – .
Wong, T. Y., Wang, J. J., Rochtchina, E., Klein, R., and Mitchell, P. (). Does refractive error influence the association of blood pressure and retinal vessel diameters? e Blue Mountains Eye Study. American Journal of Ophthalmology, (): – .
Yu, P. K., Yu, D.-Y., Alder, V. A., Seydel, U., Su, E.-N., and Cringle, S. J. (). Heterogeneous Endothelial Cell Structure Along the Porcine Retinal Microvasculature. Experimental Eye Research, (): – .
Yvonne-Tee, G. B., Rasool, A. H. G., Halim, A. S., and Rahman, A. R. A. (). Noninvasive assessment of cutaneous vascular function in vivo using capillaroscopy, plethysmography and laser-Doppler instruments: Its strengths and weaknesses. Clinical Hemorheology and Microcirculation, ():–.
Zamir, M., Medeiros, J. A., and Cunningham, T. K. (). Arterial bifurcations in the human retina. e Journal of General Physiology, ():–.
Acronyms
AASI Ambulatory Arterial Stiffness Index ...... ABPM Ambulatory Blood Pressure Monitoring ...... ACE Angiotensin Converting Enzyme ...... AECG Ambulatory Electrocardiography ...... ANOVA Analysis of Variance...... ANS Autonomic Nervous System ...... APR Average Peak Ratio ...... ARIC Atherosclerosis Risk In Communities ...... AV Atrioventricular...... AVR Arterio-Venous Ratio ...... BDF Baseline Diameter Fluctuation...... bFR Baseline-Corrected Flicker Response ...... BP Blood Pressure ...... CCD Charged-Coupled Device ...... CO Cardiac Output ...... CRA Central Retinal Artery ...... CRAE Central Retinal Artery Equivalent ...... CRV Central Retinal Vein ...... CRVE Central Retinal Vein Equivalent ...... CSV Comma-separated Values ...... CT Constriction Time ...... CV Coefficient of Variation ...... CVD Cardio-Vascular Disease ...... DA Dilation Amplitude ...... DBP Diastolic Blood Pressure ...... DD Disk Diameter ...... ECG Electrocardiogram...... FMD Flow-Mediated Dilation......
Acronyms
HRV Heart Rate Variability ...... HF High Frequency ...... HR Heart Rate ...... HT Hypertension ...... ICC Intraclass Correlation Coefficient ...... IOP Intraocular Pressure ...... IQR Inter-artile Range ...... LDR Length-to-Diameter Ratio ...... LED Light-emiing Diode ...... LF Low Frequency ...... MABP Mean Arterial Blood Pressure ...... MC Maximum Constriction ...... MD Maximum Dilation ...... MU Measuring Units ...... NN Normal-to-Normal ...... NO Nitric Oxide ...... OD Optical Density ...... ODR Optical Density Ratio ...... ONH Optic Nerve Head ...... OPP Ocular Perfusion Pressure ...... PNS Parasympathetic Nervous System...... PP Pulse Pressure ...... PSD Power Spectral Density ...... RT Reaction Time ...... RVA Retinal Vessel Analyser ...... SA Sinoatrial ...... SBP Systolic Blood Pressure ...... SD Standard Deviation ...... SITA Swedish Interactive resholding Algorithm ...... SNS Sympathetic Nervous System ...... OSat Oxygen Saturation ...... SV Stroke Volume ...... SVR Systemic Vascular Resistance ...... TIFF Tagged Image File Format...... VF Visual Field ...... VLF Very Low Frequency ......
Page removed for copyright restrictions.